UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Modelling the trophodynamics of a coastal upwelling system Robinson, Clifford L. K. 1994

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1994-893656.pdf [ 7.45MB ]
Metadata
JSON: 831-1.0053479.json
JSON-LD: 831-1.0053479-ld.json
RDF/XML (Pretty): 831-1.0053479-rdf.xml
RDF/JSON: 831-1.0053479-rdf.json
Turtle: 831-1.0053479-turtle.txt
N-Triples: 831-1.0053479-rdf-ntriples.txt
Original Record: 831-1.0053479-source.json
Full Text
831-1.0053479-fulltext.txt
Citation
831-1.0053479.ris

Full Text

MODELLING THE TROPHODYNAMICS OF A COASTAL UPWELLING SYSTEM by CLIFFORD LEIGH KELLOW ROBINSON B . S c , The University of Victoria, 1985 M.Sc , The University of Alberta, 1988 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Department of Oceanography) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA April 1994 © C.L.K. Robinson, 1994 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. (Signature) Department of ( J?&/^ <P< ^ ?/p/c( / The University of British Columbia Vancouver, Canada Date ^On/ ^7^/9V DE-6 (2/88) ABSTRACT Climate variability and its influence on the northeastern Pacific Ocean is of concern to fisheries oceanographers because of the potential impacts on fish production, and management implications about fishery potential. The main objective of my dissertation is to evaluate the interactions between oceanic variability, plankton production, and fish production in a coastal upwelling system off southwestern British Columbia, Canada. A simulation model is constructed that describes the feeding interactions among diatoms, copepods, euphausiids, juvenile and adult Pacific herring (Clupea harengus pallasi), Pacific hake (Merluccius productus), chinook salmon (Oncorhynchus tshawytscha), and spiny dogfish (Squalus acanthias). The trophodynamics model is forced by empirical seasonal patterns in upwelling, sea surface temperature, solar radiation, and by observed hake and herring biomass. The most important simulation result is that there is an imbalance between fish consumption and euphausiid production during the summer upwelling season. This highlights the requirement for improved spring estimates of zooplankton biomass, evidence of euphausiid import and aggregation mechanisms, and basic life history information for the dominant euphausiid Thysanoessa spinifera. Using observed environmental data and hake and herring biomasses from 1972 to 1990, the model estimates primary production to range from 250-500 g C m y , secondary from 5-50 g C m y , and tertiary from 0.5-4.5g C m y"1. Interannual and longer-term variability in plankton production occurred, with some years favouring copepod versus euphausiid production and vice versa. The trends in plankton production are determined primarily by variability in the dynamics of spring and summer upwelling. The average annual transfer efficiency (TE) of production from diatoms-to-zooplankton ranged 5-14% during 1972-90, while the 19 y average seasonal TE ranged 0-70%. Oceanic factors strongly influence euphausiid production, and ultimately the transfer of secondary production to hake and herring. Simulations also indicate that hake predation only significantly effects zooplankton production, while upwelling rate determines production at all trophic levels, from diatoms to fish. This research provides insights into how the productivity of a coastal upwelling region may respond to temporal variability in ocean climate. ii TABLE OF CONTENTS Abstract ii Table of Contents iii List of Tables v List of Figures viii Acknowledgements xiii General Introduction 1 La Perouse Bank 5 Objective and Approach 6 Dissertation outline 8 Chapter 1: Modelling the Trophodynamics of a Coastal Upwelling System 10 1.1 Introduction 10 1.2 Juan de Fuca Eddy feeding pathways 11 1.3 Physical Domain of the model 12 1.4 Environmental forcing functions 14 1.4.1 Upwelling 14 1.4.2 Sea surface temperature 18 1.4.3 Bright Sunshine 18 1.5 Structure of the trophodynamics model 22 1.5.1 Diatom biomass dynamics 22 1.5.2 Copepod biomass dynamics 29 1.5.3 Euphausiid biomass dynamics 36 1.5.4 Hake biomass dynamics 39 1.5.5 Pacific Herring biomass dynamics 44 1.5.6 Biomass dynamics of spiny dogfish and chinook salmon 46 Chapter 2: Calibration, Corroboration, and Sensitivity of the Juan De Fuca Eddy Trophodynamics Model 50 2.1 Introduction 50 2.2 Model calibration 50 2.2.1 Diatom biomass pattern 50 2.2.2 Copepod biomass pattern 51 2.2.3 Euphausiid biomass pattern 54 2.2.4 Fish biomass patterns 71 2.3 Model corroboration 74 2.3.1 Production dynamics 76 2.3.2 Transfer efficiency 85 2.3.3 Annual mass balance budget 87 2.4 Model sensitivity 92 2.4.1 Starting concentrations of state variables 93 2.4.2 Parameter perturbations 98 2.4.3 Variability in hake migration 103 i n Chapter 3: The Influence of Ocean Climate on Coastal Plankton and Fish Production 110 3.1 Introduction 110 3.2 Methods 112 3.3 Results 113 3.3.1 Interannual simulations 115 3.3.2 Effect of factors on 1972-90 production estimates 120 3.3.3 Comparison of model output to empirical data 125 3.3.4 Relationships between production and ocean climate 128 3.4 Discussion 135 3.4.1 Model limitations on production estimates 135 3.4.2 Interannual trends in plankton production 138 3.4.3 Long-term trends in plankton production 142 3.4.4 Fish Production 144 Chapter 4: Interactions Between Resource Limitation and Predation in a Coastal Up welling Food-web 152 4.1 Introduction 152 4.2 Methods 154 4.3 Results and Discussion 155 4.3.1 Bottom-up interactions 157 4.3.2 Top-down interactions 167 4.3.3 Temporal variability in BU/TD interactions 173 General Discussion 181 The trophodynamics model 181 Evaluation of model assumptions and structure 182 Temporal variability in production 183 Bottom-up versus top-down processes 184 Overview 185 Future directions 186 Literature Cited 188 i v LIST OF TABLES Table 1.1. Table 1.2. Table 1.3. Table 2.1. Table 2.2. Table 2.3. Summary of diatom parameters and environmental constants used in the model and discussed in the diatom biomass dynamics section (section 1.5.1; m: metres). Summary of zooplankton parameters used in the model and discussed in the copepod biomass dynamics section (1.5.2), and the euphausiid biomass dynamics section (1.5.3) (DW: dry weight; |ig C: micrograms carbon; BWD: body weight per day). Summary of fish parameters used in the model and discussed in the hake, herring, dogfish, and salmon biomass dynamics sections (1.5.4, 1.5.5, 1.5.6, respectively) (kt: thousands of tonnes; BWD: body weight per day; DW: dry weight). Observed maximum and average differences in night versus day catches of euphausiids using various sampling gear. In each case night biomass estimates of euphausiids exceed day estimates by the amount indicated. Gear abbreviations are: BN: bongo net; PN: plankton net; MN: MOCHNESS; PL: plummet net; BI; BIONESS. Maximum euphausiid lengths: Euphausia pacifica 25 mm; Nematoscelis megalops 25 mm; Thysanoessa longipes 30 mm; T. gregaria 30 mm; Meganyctiphanes norvegica 31 mm; T. spinifera 34 mm. (N: night, D: day, X: times). Estimates of zooplankton biomass (mg DW m"3) collected by Bongo nets in the Eddy and Shelf regions, during night and day (AV: average; SD: standard deviation; N: number of samples; N/D: night/day ratio). Measured maximum growth rates for adults of four species of euphausiids (See references). Growth in length is converted to growth in weight using the appropriate length-weight regressions. Also, estimated growth is calculated by assuming that starting length for each species is 12 mm, and that the euphausiids grew maximally for 30 days. References for Euphausia pacifica: Heath (1977), Bollens et al. (1992); Euphausia lucens: Stuart and Pillar 1986), Stuart 1986); Nctyphinanes australis: Ritz and Hosie (1982); Thysanoessa spinifera: Fulton et al. (1982), R. Tanasichuk, DFO, pers. comm. (L: length in mm; W is weight as mg wet weight for E. pacifica and T. spinifera, and as mg DW for N. australis and E. lucens). Page 28 35 49 58 60 68 v Table 2.4. Table 2.5. Table 2.6. Table 2.7. Table 2.8. Table 2.9. Table 2.10. Table 3.1. Simulated annual plankton and fish production properties 77 for the Eddy region during the period 1985-89. Transfer efficiency is calculated by dividing zooplankton or fish production by diatom production and multiplying by 100. Adult herring properties are calculated for 1 April to 15 November, while hake properties are for 1 June to 15 October (Cop: copepods; Euph: euphausiids). Published euphausiid production to mean biomass (P/B) 82 ratios. Published estimates of phytoplankton (1°), zooplankton (2°), 84 and fish (3°) production in productive coastal non-upwelling and upwelling regions. Transfer efficiencies for diatoms-to-zooplankton ( T E ^ ) , and for diatoms-to-fish ( T E 1 3 ) are calculated. Percentage change in annual plankton and fish production 94 from standard run values (see Table 2.4), using listed starting values of state variables. (Changes < 1.0% are denoted as 0; kt: thousands of tonnes; DW: dry weight). Standard run log-likelihood "f" values for copepod and 95 euphausiid simulations versus mean "f" values obtained by perturbing starting concentrations of state variables, and by perturbing abiotic, plankton, and fish parameters (see Table 2.9). General groupings of parameters evaluated in the sensitivity 99 analyses. Simulations are conducted by perturbing one parameter at a time by plus or minus 10% of its nominal values (See Tables 1.1, 1.2, 1.3). Coefficients of variation (%) in annual production and 100 mean annual biomass for diatoms, copepods, euphausiids, adult herring, and hake resulting from individually perturbing parameters in the general groups by plus or minus 10% of nominal values (see Table 2.9). Values of fish biomass and catch used in the Eddy model 114 simulations from 1972 to 1990. Hake biomass is from a SST/biomass relationship derived by Ware and McFarlane (1994); hake catch is from Leaman (1992); adult herring biomass is from Schweigert et al. (1992); juvenile herring biomass is calculated from adult herring biomass (see text). All biomasses and catch are in thousands of tonnes. Table 3.2. Correlations between standard run plankton production and plankton production produced using mean 1972-90 Ekman transport, or sunshine, or sea surface temperature, or hake biomass patterns. The higher the correlation the more similar the interannual plankton production pattern is to the standard run patterns (See Figures 3.1, 3.2, 3.3). Note that using a mean Ekman pattern produces the production patterns with the least similarity to standard run patterns. 126 VI Table 3.3. Table 3.4. Table 4.1. Table 4.2. Table 4.3. Table 4.4. Correlations among seasonal plankton production and 130 corresponding seasonal abiotic anomalies (Sprg = March, April, May; Sum = June, July, August; Fall = September, October; ET = Ekman transport, SST = sea surface temperature, BS = bright sunshine). Note that lagged correlations between the current season's plankton production and the previous season's abiotic anomalies are included. For N=19, the correlation coefficient has to exceed ± 0.575 and ± 0.456 to be significant at the 1% and 5% level, respectively. Classification of years by anomalies in Ekman transport 134 (ET), sea surface temperature (SST), bright sunshine (BS) (where + is positive, - is negative, L is low, M is moderate, H is high, and +/- is variable; refer to Figure 3.11). Correlations between annual plankton and fish production 158 generated from the 1972-90 simulation (See Chapter 3). For N=19, the correlation coefficient has to exceed ± 0.575 and ± 0.456 to be significant at the 1% and 5% level, respectively. Regression statistics for mean annual plankton and fish 161 production plotted against gradients in hake biomass or Ekman transport anomalies for A) the 1972-90 simulations and B) the model experiments (See Figures 4.2 and 4.5; r: correlation coefficient; CV: coefficient of variation; b: regression slope). Mean absolute percent deviation in annual plankton and 164 fish production from standard run values (see Table 2.4), in response to perturbations in model parameters belonging to major groups discussed in Table 2.9. Interaction #s refer to those included in Figure 4.1. Correlations between prey-resource regression residuals and 176 predator, for plankton biomass and production estimated in each year during the 1972-90 simulation. For N=19, the correlation coefficient has to exceed ± 0.575 and ± 0.456 to be significant at the 1% and 5% level, respectively. vn LIST OF FIGURES Figure 1 Figure 2 Figure 1.1. Figure 1.2. Figure 1.3. Figure 1.4. Figure 1.5. Figure 1.6. Figure 1.7. Figure 1.8. Figure 2.1. Figure 2.2. Figure 2.3. Catches of Pacific herring (solid line), Pacific sardines (dashed line), and Pacific hake (dots), off southwestern Vancouver Island, British Columbia, since the early 1920s (Unpublished data from D. Ware). Location of the Juan de Fuca Eddy within the northern Coastal Upwelling Domain. Arrows represent the generalized direction of surface currents around the continental shelf (< 200 m) during the April to October upwelling season (After Thomson et al. 1989; Mackas 1992). Overview of the feeding interactions (solid lines and arrows) and other processes included in the Juan de Fuca trophodynamics model. Water temperature effects diatom growth, all feeding interactions, and hake biomass. Average weekly Ekman transport at 48°N and 125°W during 1985-89. Spring transition estimated from weekly Ekman transport data at 48°N versus transitions estimated from sea level data at 42°N (Hollowed 1990). Average weekly sea surface temperature (SST) pattern at Amphitrite Point during 1985-89. Comparison of bi-monthly averaged SST at Amphitrite Pt. and bi-monthly SST from objective mapping for the La Perouse region (from R. Thomson, La Perouse 1990 Annual Report). Average weekly pattern of hours of bright sunshine measured at Tofino (49°N, 125°W) during 1985-89. Function used to describe the relationship between zooplankton growth efficiency (GGE) and changing prey concentrations. Proportion of euphausiids and herring in the hake ration plotted against estimated euphausiid biomass for June to September (See Chapter 2). Simulated weekly averaged diatom biomass pattern for the period 1985-89. Location of sampling stations used to estimate the seasonal zooplankton biomass pattern for the Eddy region. Simulated weekly averaged copepod biomass patten calibrated to the 1985-89 empirical data. Boxes represent plus or minus one standard error around the mean sampling time and observed concentration. Page 2 13 15 17 19 20 21 31 41 52 53 55 V l l l Figure 2.4. Figure 2.5. Figure 2.6. Figure 2.7. Figure 2.8. Figure 2.9. Figure 2.10. Figure 2.11. Figure 2.12. Figure 2.13. Figure 2.14. The observed biomass patterns for T. spinifera and E. 56 pacifica for the Eddy region during 1985-89. The total number of samples collected (and daylight samples) are indicated. A * represents collections 1 h after sunrise and 1 h before sunset (daytime), while small boxes are collections 1 h after sunset and 1 h before sunrise (night-time). Size frequency of T. spinifera caught using an Isaac-Kidd 64 midwater trawl at night off southwestern Vancouver Island and northern Washington (from Day 1971; model lengths are 21 mm in May and 14-15 mm in November). The seasonal biomass patterns of T. spinifera and E. pacifica 65 after correcting day samples (stars) by 20 and 2.5 times, respectively. Night- t ime and crepuscular samples (squares) are not corrected. Lines joins arithmetic averages. Note that 1 sample from T. spinifera data is excluded from October mean because it is a statistical outlier. Also note that 23 of 50 samples are collected during the day. Refer to Figure 2.4. Comparison of the standard run euphausiid biomass pattern 69 (See Figure 2.8) with patterns generated using different euphausiid growth efficiencies (20%, 25%, 35%). Boxes indicate one standard error in sampling time and concentration. Simulated euphausiid biomass pattern calibrated to 1985-89 72 empirical data. Boxes indicate one standard error in sampling time and concentration. Simulated annual biomass patterns of juvenile herring, 73 adult herring, and Pacific hake for the Eddy region during 1985-89. Calibration of simulated hake ration (A) and fraction of 75 euphausiids in hake diet (B) to empirical data (X's) from Tanasichuk et al. (1991). Simulated monthly production to mean biomass ratios (P/B) 80 for diatoms, copepods, and euphausiids. Zooplankton production (solid line and squares) and 86 transfer efficiency (broken line and triangles) plotted against annual primary production. Panel A: data from Cushing 1971. Panel B: data from Table 2.6. Fish production (solid line and squares) and transfer 88 efficiency (broken line and triangles) plotted against annual primary production. See Table 2.6 for values. Flow diagram of the standard 1985-89 model run. Flows are expressed as tonnes km y . Note that the size of the plankton boxes are proportional to the natural logarithm of their annual production (P: production; B: mean annual biomass). 89 IX Figure 2.15. Figure 2.16. Figure 2.17. Figure 2.18. Figure 2.19. Figure 2.20. Figure 2.21. Figure 3.1. Figure 3.2. Figure 3.3. Figure 3.4. Figure 3.5. Figure 3.6. Annual plankton production produced over 50 consecutive 97 simulations. Panel A: Diatom and euphausiid production with +/- 5% first year boundaries. Panel B: Diatom and copepod production showing slight oscillations over time. Response of simulated diatoms to higher (4.4 m d"1) and 101 lower (3.6 m d"1) upwelling rates compared to the standard run (4.0 m d ' 1 ) . Response of simulated copepods to higher (44%) and lower 102 (36%) growth efficiency compared to the standard run (40%). Response of simulated euphausiids to early (dots) and late 104 (dashes) hake arrival, compared to the standard run (line). Simulated effects of variable hake immigration date on 106 annual plankton production. Changes in production are calculated as % deviations from the standard run values listed in Table 2.4. Response of model euphausiids to higher (dashed line; 315 107 kt) and lower (dots; 105 kt) hake biomass, compared to the standard run (line). Simulated effect of variable hake biomass on annual 109 plankton production. Changes in production are calculated as % deviations from standard run values listed in Table 2.4. Simulated estimates of annual diatom production for the 116 Juan de Fuca Eddy from 1972 to 1990. The range in production in a given year (bars) is determined by using variable starting plankton concentrations for 1972 (see text). The long-term trend (dashed line) is determined using a LOWESS smoother (tension = 0.5). Simulated estimate of annual copepod production for the 117 Juan de Fuca Eddy from 1972 to 1990. See Figure 3.1 legend. Simulated estimate of annual euphausiid production for the 118 Juan de Fuca Eddy from 1972 to 1990. See Figure 3.1 legend. Simulated estimates of annual herring and hake production 119 for the Juan de Fuca Eddy from 1972 to 1990. The long-term trends (dashed lines) are determined using a LOWESS smoother (tension = 0.5). Comparison of diatom and copepod production from the 122 standard continuous simulation (solid lines) to production generated from running each model year separately. Comparison of standard run diatom and copepod annual 123 production (solid lines) to simulations without euphausiid import (dots), and to simulations with lowered copepod offshore transport (dashed lines). x Figure 3.7. Comparison of standard run euphausiid and herring annual 124 production (solid lines) to simulations without euphausiid import (dots), and to simulations with lowered copepod offshore transport (dashed lines). Figure 3.8. Comparison of annual diatom and copepod production from 127 the standard run (solid lines) to production from a continuous simulation using the 1972-90 mean Ekman transport pattern (dashed lines). Figure 3.9. Panel A: Condition of age 5 herring from 1972 to 1990. 129 Panel B: Relationship between herring condition and simulated autumn zooplankton production (r= +0.58). Figure 3.10. Panel A: Interannual pattern in spring (dashed line) versus 132 summer (solid line) Ekman transport anomalies (from 1972-90 mean). Panel B: Interannual variability in the day of the spring transition using Ekman data for 48°N compared to sea level data from 42°N (data from Hollowed 1989). Figure 3.11. Grouping of years by similar abiotic anomalies measured 133 during the upwelling season for 1972-90. Groups determined using the Euclidean distance measure and average linkage clustering technique (Wilkinson 1990). Numbers above groups refer to those listed in Table 3.4. Figure 3.12. Relationship between simulated spring (March-May) 139 copepod biomass and that measured at Ocean Weather Station P during 1967-80 (data from McFarlane and Beamish 1993; r= -0.34). Figure 3.13. Relationship between euphausiid summer production and 146 observed herring age 5 condition factor for the period 1972-89 (unpublished data from D. Ware; r= -0.33). Figure 3.14. The interannual pattern in the efficiency of transfer from 148 diatoms to zooplankton (euphausiids and copepods) and from diatoms to fish (hake and herring). Transfer efficiency is calculated by dividing fish or zooplankton production by diatom production and multiplying by 100. Figure 3.15. Monthly diatom to zooplankton transfer efficiency and 149 diatom production averaged over 1972-90. The bars represent the 95% confidence interval around the mean transfer efficiencies. Figure 3.16. Coefficient of variation (CV) in Ekman transport averaged 151 weekly from 1972 to 1990. Figure 4.1. Conceptual overview of the relative strength of "bottom-up" 156 versus "top-down" influences on trophic level biomass in the present-day Juan de Fuca upwelling system (thick solid lines: strong effect; thin solid line: moderate effect; thin dashed lines: weak effect; odd numbers refer to bottom-up processes, while even numbers refer to top-down processes discussed in the text). xi Figure 4.2. Annual plankton production plotted against upwelling rate 159 (Panel A) and against observed Ekman transport (Panel B). Figure 4.3. Annual copepod (solid line) and euphausiid (dashed line) 162 production plotted against diatom production (r= +0.09, and r= +0.27, respectively). Figure 4.4. Annual herring and hake production from 1972-90 166 simulation plotted against annual euphausiid production (r= +0.11, r= +0.93, respectively). Figure 4.5. Simulated annual plankton production plotted against 168 relative (A) and estimated (B) hake biomass. Figure 4.6. Annual copepod production plotted against annual 171 euphausiid production (r= -0.71). Data from the 1972-90 simulation. Figure 4.7. Panel A: The coefficient of variation (CV) in monthly 174 diatom biomass generated in the upwelling rate and hake model experiments. Panel B: The ratio of CVs from experiments for both diatom biomass and production. Figure 4.8. Panel A: The coefficient of variation (CV) in monthly 177 euphausiid biomass generated in the hake biomass and upwelling rate experiments. Panel B: The ratio of CVs from model experiments for both euphausiid biomass and production. Figure 4.9. Panel A: The coefficient of variation (CV) in monthly 179 herring biomass generated in the hake biomass and upwelling rate experiments. Panel B: The ratio of CVs from model experiments for herring biomass and production. xn ACKNOWLEDGEMENTS Many people deserve my heartfelt thanks for their contributions to the four years of research that culminates in this dissertation. First and foremost, I extend my deepest gratitude to my co-supervisor Dr. Daniel Ware (Fisheries and Oceans Canada) for his seemingly constant support, guidance, and enthusiasm in developing my ideas, and the simulation model. Dan has also afforded me with many opportunities to participate in the forefront of the new science of Fisheries Oceanography, for which I am very grateful. My other co-supervisor, Dr. Timothy Parsons (UBC, Oceanography), deserves thanks for developing many ideas concerning marine plankton ecology, and for the challenges provided on the tennis court; I have the distinction of being Tim's last graduate student of what truly has been an illustrious career. Several current and former researchers at the Pacific Biological Station have contributed in numerous and important ways through discussions, use of data, and advice, and deserve my thanks, these include: Dr. David Mackas, Mr. Ron Tanasichuk, Dr. Rick Brodeur, Mr. Dan Bouillon, Dr. Mark Johannes, and members of the La Perouse annual meetings. I also thank members of my supervisor committee, Drs. Dan Ware, Tim Parsons, Dave Mackas, Paul Harrison, John Fyfe, and Susan Allen for their discussions and critical, constructive reviews. Several people have been instrumental in providing "logistic" support for this work: Dr. Bill Silvert for instruction on the art of modelling with BSIM; Ms. Pam Olson and Mr. Gordon Miller for what must have seemed like constant help in the PBS library; Cris Mewis of the Oceanography Department for guiding me through the minefields of UBC bureaucracy; and Barbara Rokeby at UBC Oceanography for keeping Tim and I organized. I would also like to acknowledge the support that I have received throughout my Ph.D. studies from the Natural Sciences and Engineering Research Council of Canada, the Science Council of British Columbia, and the Fisheries and Oceans Canada. Finally, thanks Debra for your support and understanding for what must truly seem to you to be a "labour of love". x i n 1 GENERAL INTRODUCTION The most productive fishing region along the west coast of Canada is located on the continental shelf off Vancouver Island (Ware and McFarlane 1989). For the past 70 years, the majority of fish catch from this region has consisted of pelagic species like Pacific herring (Clupea harengus pallasi), Pacific sardine (Sardinops sagax), and more recently Pacific hake (Merluccius productus; Figure 1). Inherent in each time series is short-term and long-term variability in the catch. Short-term variability in the catch record may reflect real changes in the fishery. The decline in herring catches during the late 1920s, for example, can be associated with the collapse of dry-salt markets in China. Similarly, the rapid increase in herring catches in the early 1940s is partly due to the bui ld-up of the reduction fishery (Hourston and Haegele 1980). It is recognized however, that longer-term fluctuations in the catch are more likely a reflection of real changes in fish populations. Soutar and Isaacs (1969), and more recently Baumgartner et al. (1992), have determined that populations of several fish species, like Pacific sardine and Pacific hake, exhibit low frequency oscillations in biomass over several hundreds of years. Since the nature of these long-term oscillations exclude the impact of fishing, the majority of recent research has focused on relationships involving causes of natural variability. It is generally thought that longer-term variability in fish populations is linked to alterations in fish recruitment (e.g., Hollowed 1990), or migrations (e.g., Ware and McFarlane 1994), among other factors, associated with a variable oceanic environment or "ocean climate". To understand the influence of ocean climate on temporal variability in fish populations found off Vancouver Island, it is important to recognize that this shelf region is located at the northern terminus of the Coastal Upwelling Domain; the Domain 2 50 Hake 100 o ° o^ > n.Q ^ ^ ^ «.Q <$> c ° e$> A ° ^ «.° o^ > oP & & & &> N0* #* & N<£ & fP & & ^ ^ ^ CO o c c O O *-> a o a> cs X Year Figure 1. Catches of Pacific herring (solid line), Pacific sardines (dashed line), and Pacific hake (dots), off southwestern Vancouver Island, British Columbia, since the 1920s (Unpublished data from D. Ware and B. Waddell, DFO, Nanaimo). 3 extends southward to Baja, California at 25°N (Figure 2; Ware and McFarlane 1989). Note that the Coastal Upwelling Domain is also known as the California Current system; I will use the former in my dissertation. The Coastal Upwelling Domain is an eastern boundary current system that consists of three major oceanic currents. The slow, equatorward flowing California current is strongest during summer, and in the northern Domain it appears to be well seaward of the continental shelf. The jet- l ike California Undercurrent is a subsurface poleward flow generally confined to the continental slope. During winter in the northern Domain, a poleward surface flow called the Davidson Current occurs over the continental shelf (Hickey 1979). One of the most important physical features of this eastern boundary current system is the seasonality in alongshore wind stress. Equatorward (northwesterly) winds result in offshore transport of nearshore surface waters, and subsequent upwelling of relatively deep, cool nutrient rich waters to the surface layer. Upwelling occurs year round in the Domain south of about 40° N, while occurring only seasonally further north (Hickey 1979; Huyer 1983). In the northern Domain, a seasonal shift from downwelling to upwelling favourable winds takes place in March or April as a result of the intensification of the North Pacific High pressure system and a weakening of the Aleutian Low pressure system (Strub and James 1988). Conversely, in the autumn a strengthening of the Aleutian Low pressure system results in a shift from northwesterly to southeasterly alongshore winds and subsequently downwelling conditions (Huyer 1983; Strub and James 1988). The seasonal variability in the alongshore winds in the northern Domain has at least two important consequences for fishes. The biomass and production of zooplanktonic prey is linked to seasonal patterns and interannual variability in oceanic conditions, like upwelling (e.g., Peterson and Miller 1975; Mackas 1994). In addition, the biomass of fishes occurring in the northern Domain is largely determined by temporal variability in water currents or temperature acting on migratory species (Bailey et al. 1982; Ware and 125° 115° British Columbia Washington Oregon 50° Shelf-Break Current Region / ^ZZ 40° 30° 126° 125° FIGURE 1. Location of the Juan de Fuca Eddy within the northern Coastal Upwelling Domain. Arrows represent generalized direction of surface currents around the British Columbia continental shelf (< 200m) during the April to October upwelling season (after Thomson et al. 1989 and Mackas 1992). 5 McFarlane 1994). In fact, the migratory range of the Pacific hake, and historically the Pacific sardine, delimits the northern extent of the Coastal Upwelling Domain (Ware and McFarlane 1989). La Perouse Bank Given that fishes occurring in the northern Domain are influenced by large-scale, seasonal oceanic processes, it is also necessary to recognize that oceanographic properties off Vancouver Island may influence population variability. The majority of the present-day fishing activity occurring along the west coast of Vancouver Island centres around La Perouse Bank, named after the French explorer who travelled along the BC coast in August 1786. It is interesting to note that La Perouse had attempted to land at Friendly Cove in Nootka Sound, but could not because of thick coastal fog (Shelton 1987). The conditions experienced by La Perouse over 200 years ago are typical of BC coastal upwelling conditions today. During the April to October upwelling season, the waters around La Perouse Bank are defined by three relatively distinct physical and biological subregions (Figure 2; Thomson and Ware 1988; Mackas 1992). The subregion furthest seaward, the shelf break and slope, is dominated by the southeastward flowing Shelf Break Current and consists of relatively warm, saline water. Nutrient and phytoplankton concentrations in this region are lowest of the three subregions, while zooplankton biomass has the least seasonality and is dominated by oceanic calanoid copepods, chaetognaths, and salps (Mackas 1992). The inner shelf subregion, adjacent to Vancouver Island, is dominated by the northwest flowing Vancouver Island Coastal Current. These waters are cool, low in salinity, and originate from Juan de Fuca Strait (Thomson and Ware 1988). Surface waters of the inner shelf typically contain high nutrient and phytoplankton concentrations but consistently low average biomasses of zooplankton dominated by neritic calanoid copepods and jellyfish. Compared to the other two subregions, chaetognaths and euphausiids make 6 small contributions to the total zooplankton biomass (Mackas 1992). The subregion east of Juan de Fuca Strait and intermediate of the first two subregions, is characterized by a quasi-permanent cyclonic gyre, frequently called the Tully Eddy or the Juan De Fuca Eddy; the latter is used in this study. The Juan de Fuca Eddy is formed in March/Apri l and lasts until September/October, primarily because of upwelling currents (Freeland and Denman 1982). Of the three oceanic subregions, the Juan de Fuca Eddy contains the highest average seasonal concentrations of zooplankton, particularly large calanoid copepods and euphausiids; conversely, gelatinous zooplankton biomass is of lesser importance in this oceanic subregion (Simard and Mackas 1989; Mackas 1992). In addition, the Eddy region contains the largest concentrations of migratory fishes like Pacific hake, as well as resident species like Pacific herring, spiny dogfish (Squalus acanthius), and Chinook salmon {Oncorhynchus tshawytscha) (Tanasichuk et al. 1991; Ware and McFarlane 1994). In 1985 the Canadian Department of Fisheries and Oceans (DFO) at the Pacific Biological Station in Nanaimo BC, and the Institute of Ocean Sciences in Sidney BC, initiated a long-term, multi-disciplinary oceanographic study to improve the management of fish stocks in the La Perouse region. The La Perouse Project is designed to identify the dominant physical factors effecting general water properties and circulation, to quantify the seasonal and interannual variability in plankton biomass, and to understand the feeding interactions between dominant plankton and pelagic fishes (Ware and Thomson 1986). Objective and Approach The main objective of my dissertation is to use information from DFO's initiative to provide a holistic view of the temporal production dynamics of pelagic plankton and fish, in the present-day La Perouse ecosystem. To accomplish this objective, I have developed a micro-computer simulation model describing the feeding interactions of the 7 most abundant pelagic organisms found in the Juan de Fuca Eddy subregion. Production dynamics are modelled for this region because of the large biomasses of meso/macro zooplankton and fish (see above). I used a simulation modelling approach because it synthesizes available knowledge, evaluates multi-factor interactions, and identifies important data gaps (Silvert 1981). The trophodynamic approach estimates lower bounds of production from available data describing standing stocks and feeding requirements of dominant pelagic organisms (Laevastu et al. 1981). Prey biomasses and production can also be related to changes in other prey and predators (i.e, allows for multi-species effects), and to non-trophodynamical losses. To evaluate the influence of oceanic variability on production dynamics, I have forced the trophodynamics simulation model with empirical seasonal patterns in upwelling, sea surface temperature, and solar radiation. Relatively few trophodynamical models exist for coastal upwelling systems, compared to other productive coastal regions (e.g., Norths Sea; Franz et al. 1991), and even fewer trophodynamic models consider feeding interactions from plankton to fish. The majority of simulation models developed for coastal upwelling regions have been primarily restricted to the Humbolt and Benguela Current systems. Walsh (1975) for example, developed a spatial simulation model of the Peru upwelling system to investigate the production dynamics of anchoveta. Jarre et al. (1990) have also examined "steady-state" energy flows from plankton to fish, for three different periods, in the Peru upwelling system. Several recent modelling studies have described energy flows in the Benguela upwelling system (see Moloney 1992). However, these studies have ignored plankton/fish trophodynamics, and have considered a constant coastal upwelling environment. In the Coastal Upwelling Domain in general, and in the northern Domain in particular, relatively few studies have modelled any pelagic trophodynamic interactions. In fact, during the past three decades, < 12 studies have been published. In the northern Coastal Upwelling Domain, a couple of modelling studies have evaluated the dynamics of 8 phytoplankton blooms in Puget Sound in relation to seasonal variability in oceanic conditions, while only minimally considering zooplankton grazing dynamics (e.g., Winter et al. 1975; Jamart et al. 1977). A few studies have also focused on detailed evaluations of mesozooplankton (copepod) population dynamics off coastal Oregon, but have generally ignored their feeding interactions (e.g., Wroblewski 1980; Peterson et al. 1979). Plankton-plankton trophodynamics in the NE Pacific Ocean have been most fully evaluated in oceanic regions far removed from the productive northern coastal upwelling Domain (e.g., Frost 1987). Very few studies have attempted to evaluate the trophodynamics of coastal zooplankton and fish in the northern Domain, and these have been restricted to larval or juvenile stages (e.g., Parsons and Kessler 1986, 1987; Wroblewski and Richman 1987). Laevastu and colleagues in the 1970s modelled fish-fish and fish-zooplankton feeding interactions to investigate various scenarios of generalized zooplankton production and fishing mortality on long-term fish biomass. These modelling studies were however developed for the stratified Bering Sea. See Laevastu et al. 1982 for an overview. A couple of recent models have used particle-size theory to estimate production of, or biomass transfer between, generalized "trophic levels" in coastal BC waters (Dunbrack and Ware 1986; Denman et al. 1989). These two studies however, generally ignored changes in the phasing or timing of abiotic or biotic events that may effect the transfer of biomass from prey to predator. To sum up, Parsons (1991) notes that there is a need for modelling studies to incorporate plankton and fish interactions in combination with changes in the coastal oceanic environment in the northeastern Pacific. Dissertation outline Chapter 1 of my dissertation discusses the structure and assumptions of a microcomputer simulation model describing the trophodynamics of the most abundant 9 pelagic plankton and fishes found in the Juan de Fuca Eddy. The second Chapter has three major components. I begin by calibrating the standard model output to empirical data collected for the La Perouse region. I then corroborate the production dynamics of the model to observations made in other regions of the Coastal Upwelling Domain, and to observations made in the Benguela Current upwelling system. Finally, I present an analysis of the sensitivity of model output to perturbations in abiotic and biotic parameters, and to key processes, like hake migration. Chapter 3 of my dissertation presents the range of behaviour exhibited by the trophodynamics model to seasonal and interannual variability in the three environmental forcing functions. The fourth Chapter explores the interactions between resource and predator limitations on "trophic level" production. I conclude my dissertation with a general discussion of the major findings of this research, and the direction for future research to better understand the influence of ocean variability on the trophodynamics and production potential of a coastal upwelling system. 10 CHAPTER 1: MODELLING THE TROPHODYNAMICS OF A COASTAL UPWELLING SYSTEM 1.1 Introduction One of the major trophodynamical links between plankton and pelagic fish in coastal up welling systems occurs via net-phytoplankton to meso/macro zooplankton to fish (Ryther 1969; Parsons et al. 1984; Cushing 1989; Mann 1993). The predominance of this pathway is related to the fact that there is a relatively constant predator-to-prey size ratio in aquatic food chains (Sheldon et al. 1977). For instance, small phytoplankton (< 5 fim) are primarily eaten by protozoans, while net phytoplankton (e.g., diatoms) are primarily eaten by metazoan zooplankters, like copepods, because the latter are inefficient consumers of small particles (Parsons et al. 1984; Paffenhofer 1986; Legendre and Le Fevre 1992). There is evidence that copepods feed on top predators of the microbial loop (e.g., ciliates; Kiorboe 1991), but the transfer of biomass/energy via this pathway is relatively small (Moloney 1992), and inefficient (Cushing 1989). The diatom-metazoan pathway ultimately predominates in productive coastal systems because the net phytoplankton generally dominate phytoplankton biomass (Legendre 1990; Kiorboe 1993). Diatoms dominate phytoplankton biomass in weakly stratified regions because their relatively low ratio of respiration to maximum photosynthesis enables them to "out-compete" smaller phytoplankton (e.g., flagellates) under conditions of pulsed, high nutrient supplies and strong vertical mixing (Parsons 1979; Harrison and Turpin 1982). Kiorboe (1993) also suggests that diatoms dominate phytoplankton biomass because large cell-size offers a refuge from predation associated with lower relative densities and longer generation times of their metazoan predators. In contrast, the biomass of smaller phytoplankton remains relatively constant, in both turbulent and stratified oceanic regions, because their predators exhibit rapid numerical responses to any population increases (Parsons and Lalli 1988; Kiorboe 1991). 11 The relative dominance of phytoplankton biomass and production by diatoms in turbulent coastal areas is also dependent upon the frequency of upwelling (Wyatt 1980). A modelling study of a plankton food-web in the Benguela coastal upwelling system, for example, indicates that diatoms dominate phytoplankton biomass for about 7-10 days after an upwelling event (Moloney 1992). The microbial food-web only dominated from about day 10 to day 30 after the initial upwelling event. Physical oceanographic data from coastal regions in the California and Benguela current systems indicates that upwelling events occur roughly every 1-2 weeks (e.g., see Figure 1 in Mann 1993). Because of the relative short-term frequency of vertical mixing events one would expect diatoms to dominate phytoplankton biomass during the majority of the upwelling season. Garrison (1976) in fact observed that net phytoplankton comprised on average 80% (standard deviation; SD = 18.8) of the phytoplankton biomass during the upwelling season off Monterey, California. Legendre and Le Fevre (1992) further note that large phytoplankton cells typically dominate production as well as biomass in upwelling regions. It is the temporal (and spatial) variability in upwelling that generates diatom "blooms", and primarily determines the transfers to the meso/macro zooplankton and subsequently to fish (Wyatt 1980; Legendre 1990). I evaluate the temporal variability in coastal upwelling and its influence on the diatom-to-meso/macrozooplankton-to-fish feeding pathway in the remainder of my dissertation. 1.2 Juan de Fuca Eddy feeding pathways The relative importance of the net phytoplankton-to-meso/macro zooplankton-to -fish trophodynamical pathway in the Juan de Fuca Eddy is supported by the fact that during the upwelling season phytoplankton biomass can be dominated by diatom genera like Rhizosolenia or Nitzschia (Denman et al., 1981; Mackas and Sefton 1982). About 60-80% of meso/macro zooplankton biomass consists of large neritic calanoid copepods (e.g., Calanus marshallae), and the euphausiids Thysanoessa spinifera and Euphausia pacifica 12 (Mackas 1992). The euphausiids in turn, constitute the majority (60-80%) by weight of the diets of three fishes that make up > 80% of the upwelling season pelagic fish biomass: the Pacific hake, Pacific herring, and spiny dogfish (Squalus acanthias). C hinook salmon (Oncorhynchus tshawytscha) are also important in the region because they consume large biomasses of small fishes, like herring (Tanasichuk et al. 1991; Ware and McFarlane 1994). The remainder of Chapter 1 discusses the structure and assumptions of a microcomputer simulation model that describes the feeding interactions between diatoms, copepods, euphausiids, juvenile and adult herring, Pacific Hake, spiny dogfish, and Chinook salmon (Figure 1.1). 1.3 Physical Domain of the model Compared to the surrounding oceanographic regions on the southern BC continental shelf, the Juan de Fuca Eddy contains high concentrations of fish and plankton (Mackas 1992; Ware and McFarlane 1994). The main concentrations of fish and euphausiids occur in and around the deep basins (100-200 m) of the Eddy region, an area of about 1520 km2 . Since this a relatively large oceanic area, I have ignored detailed horizontal or vertical spatial resolution. The only generalized effect of horizontal water movement considered in the model is the export of surface dwelling plankton from the Eddy region to offshore areas. This assumption is supported by observations made by Mackas (1992) that mesozooplankton biomass, in the Juan de Fuca Eddy, generally declines and community composition changes as the upwelling season proceeds. The loss of biomass from the Eddy region to surrounding oceanic regions is due, in part, to horizontal mixing between regional gradients and to advective displacement of previously resident waters (Mackas et al. 1987; Mackas 1992). The vertical structure of the model is simply represented by a surface mixed layer homogeneous with respect to plankton and nutrients, and a lower layer containing nutrients but no plankton (after Evans and Parslow 1985; Frost 1987). During 13 _ _ _ ^ Immigration/ Immigration/ Emigration Immigration Offshore k. export 4 Upwelling (Nitrogen) Solar radiation Figure 1.1. Overview of the feeding interactions (solid lines and arrows) and other processes included in the Juan de Fuca tropho-dynamics model. Water temperature affects diatom growth, all feeding interactions, and hake biomass. 14 the upwelling season, vertical exchange is limited to upwelling of nutrients from the lower layer to the surface layer and sinking of phytoplankton from the surface layer (see section 1.5.1). Phytoplankton are not re-introduced into the surface layer by upwelling. 1.4 Environmental forcing functions The trophodynamics model (Figure 1.1; see below) is forced by empirical seasonal patterns in: upwelling, water temperature, and solar radiation, observed off southwestern Vancouver Island from 1985 to 1989. 1.4.1 Upwelling Along the west coast of North America the upwelling of relatively deep water to the surface layer results from northwesterly winds forcing near-shore waters offshore (i.e., Ekman transport; Huyer 1983; Strub and James 1988; Thomson and Ware 1988). Because few actual measurements of upwelling exist, I characterize the seasonal cycle of upwelling in the Juan de Fuca Eddy region using the coastal upwelling index of Bakun (1973). The offshore or onshore transport of coastal waters was calculated by Bakun (1973) as, surface wind stress divided by the Coriolis force. Wind stress is estimated from atmospheric pressure gradients measured at three degree latitudinal intervals along the west coast of North America. The magnitude of positive Ekman transport represents the average volume of deep water (m ) upwelled through the bottom of the Ekman layer each second along 100 m of the dominant trend in 200 miles of coastline. I derive seasonal patterns in upwelling for each year by calculating weekly averages starting 1 January of each year from daily Ekman transport (ET) data for 48°N 125°W. The Ekman data used are published by the Pacific Fisheries Environmental Group, Monterey, California. Weekly data for all 5 years are combined, and the average 1985-89 Ekman transport pattern is used to force the model (Figure 1.2). Given the seasonal pattern in upwelling, it is necessary to define the start of the 15 Upwelling Downwelling _2oo'" ' ] M ' j M l ' [ i i ii i n i ii \ \ \ \ \ \ \ \ \ \ \ \ >>»%*V* J ** ** ^ *>V* 0°V* <?* Date Figure 1.2. Average weekly Ekman transport at 48°N, 125°W during 1985-89. 16 upwelling season to initiate the upwelling of sub-surface nitrogen, and the offshore transport of surface dwelling plankton (see section 1.5.1). Strub and James (1988) indicate that the spring transition off Oregon/Washington occurs abruptly over a few days sometime in March or April. I estimate the start of the upwelling season by assuming that the spring transition occurs during the first week of two consecutive weeks of positive Ekman transport. Spring transition dates, estimated at 48°N from Ekman transport data for 1972 to 1986, are positively correlated with transition dates estimated by Hollowed (1990) using sea-level data from 42°N (Figure 1.3; r= 0.83, P < 0.01). Ekman transport also effects an additional model variable, the depth of the surface mixed layer (M). Seasonal variation in M, as defined by uniform water temperature, is assumed to be related to the intensity of Ekman transport. For example during winter, large negative Ekman values imply strong poleward winds and thus greater turbulence and downwelling resulting in a deeper mixed layer. The mixed layer depth is assumed maximal at 100 m when greatest downwelling occurs in January (Figure 1.2). M for all other weeks in winter is expressed relative to the week with the greatest negative Ekman transport. In the spring, a relatively shallow M develops in response to reduced winter winds and seasonal warming (Thomson and Ware 1988). M remains roughly 10-20 m deep throughout the summer because of upwelling and relatively low wind stress (Crawford and Dewey 1989; Thomson et al. 1989; Mackas 1992). For example, wind stress along the BC coast during the upwelling season is about 25% of that observed in winter (Figure 1.2). M is held constant at 15 m throughout the upwelling season, which is defined as the period between the spring and autumn transitions (see above). This minimum depth corresponds to the depth of the Ekman layer estimated at several coastal sites in the northern Coastal Upwelling Domain (Freeland and Denman 1982; Huyer 1983). After the autumn transition, M deepens in proportion to the pattern and intensity of negative Ekman transport to a maximum of 100 m (Figure 1.2). 17 Week of transition using Ekman transport Figure 1.3. Spring transition estimated from weekly Ekman transport data at 48oN versus transitions estimated from sea level data at 420N (Hollowed 1990). 18 1.4.2 Sea surface temperature Data describing the seasonal pattern in sea surface temperature (SST) are used to estimate diatom doubling rate, zooplankton and fish ingestion rates (Q10 effects), and the biomass of Pacific hake in the Eddy region. I assumed that sea surface temperature is descriptive of changes in temperature in the surface mixed layer (see above). The longest daily time series of SST for the southwestern coast of Vancouver Island is from Amphitrite Point light house (49°N and 127°W; Freeland 1991). Weekly SST averages are calculated from the daily SST measurements starting 1 January of each year from 1985 to 1989 (Figure 1.4). A comparison of Amphitrite Point SST with SST measured for the La Perouse Bank region (R. Thomson, La Perouse 1990 Annual report), indicates that the seasonal pattern is coherent. Sea surface temperatures are of similar magnitude in March/April and July/August, but SSTs are about 1.0 °C warmer over La Perouse Bank during the remaining months (Figure 1.5). 1.4.3 Bright Sunshine Data describing the seasonal pattern and intensity of solar radiation are required to estimate the effects of light limitation on diatom growth (see section 1.5.1). Daily data describing the hours of bright sunshine are taken from a coastal station on southwestern Vancouver Island (49°N 125°W; Environment Canada, Atmospheric Service meteorological observations). The 1985-89 weekly averaged pattern of bright sunshine at Tofino is shown in Figure 1.6. I converted hours of bright sunshine (BS) to units of solar radiation (SR in Langleys) using a linear relationship derived from data collected at the University of British Columbia: SR = 31.3 + 5.06*BS, where BS is in tenths of hours of bright sunshine (r = 0.89; N = 730). I assumed that 50% of the solar radiation reaching the sea surface is available to diatoms as photosynthetically active radiation (Parsons et al. 1984). 19 16 « i 14 2 0) (0 0) 8 a E 0) 6 « 4 0 I i i i i i i i i i i I M M I I I I I I I I i I I I I I I I I I I I I I I I N \ .0 ** «*> ^  J ^ ^ ^ ^ ^ ^ ^ J Date Figure 1.4. Average weekly sea surface temperature (SST) pattern at Amphitrite Point during 1985-89. 20 14 12 U o 10 CO (0 ^ r 'La Perouse Amphitrite Pt JF MA MJ JA SO Bi-monthly periods ND Figure 1.5. Comparison of bi-monthly averaged SST at Amphitrite Point and bi-monthly SST from objective mapping for the La Perouse region (from R. Thomson, La Perouse 1990 Annual report). 21 10 0) c £ (0 c 3 (0 a g ^A Q I I I I I I I I I I I I I I I I I I M I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I [ N \ \ N \ N \ \ \ \ \ \ * * « * # ^ * * >** ^ ^ - * ~* * "* ^ *° o° Date Figure 1.6. Average weekly pattern of hours of bright sunshine measured at Tofino (49oN, 125QW) during 1985-89. 22 1.5 Structure of the trophodynamics model This section describes the mathematical equations used to represent the major processes effecting the biomass of diatoms (D), copepods (C), euphausiids (E), adult herring (H), juvenile herring (J), Pacific hake (W), spiny dogfish (S), and chinook salmon (O). The system of difference equations are integrated forward, from empirically derived initial conditions, by 1 d time intervals for 1 y using a software package, Biological SIMulation (BSIM; Silvert 1989). Integrations shorter than 1 d are not warranted because of additional assumptions and complexities required to represent short term physiological and behavioral characteristics of the plankton (e.g., Hofmann and Ambler 1988). Diatom biomass is modelled as milligrams chlorophyll a per cubic metre (mg Chla m ), while zooplankton biomass is in grams dry weight per square metre (g DW m" ). Fish biomass is in metric tonnes of wet weight per square kilometre (1 t WW km" = 1 g WW m" ). Carbon (C) is the common unit in the model. To convert chlorophyll a to carbon, I use a ratio (Cc) that varies from 35:1 to 50:1. It is well known that Cc is lower when phytoplankton are growing vigorously compared to when populations are resource (nutrient/light) limited (Parsons et al. 1984). I assume that 15% of fish wet weight is dry weight (Wd), 45% of fish (and zooplankton) dry weight is carbon (Dc), and thus 6.75% of fish wet weight is carbon (Wc; Parsons et al. 1984)). 1.5.1 Diatom biomass dynamics On the southern BC continental shelf chlorophyll a ranges from 0.1-1.0 mg Chla m" in winter to 5-20 mg Chla m"3 in summer (Mackas and Sefton 1982; Forbes et al. 1987; Mackas 1992). I assume that chlorophyll a consists primarily of net phytoplankton (e.g., Mackas and Sefton 1982), and use a starting concentration of 0.5 mg Chla m" . The processes included in the model that effect diatom biomass are described by the difference equation: 23 Equation 1: Dt+1 = Dt + A/) / UdNLD A* UN+H,)(L + H) Cc - TdTeDCc - M ( l - | - ) D C ( (ZIZ(D-Pt) yH*+D-p„ D. = diatom growth - offshore export of diatoms - diatom sinking - zooplankton predation Where D is diatom biomass, U d is the maximum diatom doubling rate, N is the mixed layer nitrogen concentration, L is the mixed layer light, H n and Hj are Michaelis-Menten half-saturation constants, Rd is daily diatom respiration, Cc is chlorophyll a to carbon conversion factor, Td is the maximum daily export rate of diatoms, Te is the seasonal Ekman transport pattern, Sd is the maximum diatom sinking rate, M is the mixed layer depth, Z is copepod and euphausiid biomass, Iz is the zooplankton ingestion rate, H z is the grazing half-saturation constant, P t is the zooplankton feeding threshold, and D c is the dry weight to carbon conversion factor. Values of these parameters are discussed below and summarized in Table 1.1. Diatom growth: The first term of the diatom biomass equation represents the temperature dependent doubling rate (Ud) limited by the effects of light (L) and nitrogen (N) concentrations. Daily diatom respiration (Rd) is also considered in equation 1. Phytoplankton typically respire about 5-15% of their daily gross production; I use a mid-range value of 10% (Parsons et al. 1984).The maximum doubling rate of diatoms (Ud) is regulated by water temperature (Eppley 1972). Because Eppley's relationship is derived from phytoplankton cultures grown under continuous light, I adjusted for hours of daylight (DL) using a cosine function varying between a maximum of 16.2 h on 21 June, and a minimum of 8.1 h on 21 December for 50°N (Nautical Atlas). From Frost (1987), maximum doubling rate can be defined as: 24 Equation la: Ud = (e u l n z - 1.0) DL/24 where, Equation lb: D = doublingsd'1 = 0.851(10)' \0.0275 *TEMP Equation \c: DL = daykngthih) = 12.25 - 4.15(COS2it(DAF-l)/360) The maximum diatom doubling rate (Ud) is ultimately limited by mixed layer nitrogen concentrations and average light using Michaelis-Menten dynamics (see equation 1; after Parsons and Takahashi 1973). Nitrogen dynamics in the surface mixed layer can be represented by the following equation: Equation 2: Nt+1 = Nt + AN At U„NLD \ AN At UN + H,)(L+H} ^ + I^ J + (XzIzZ(P-Pt)\ \(Hp+P-P) N, diatom uptake + upwelling + zooplankton excretion Where N is the nitrogen concentration, U is the maximum upwelling rate, N 3 0 is the nitrogen concentration below the mixed layer, and X z is the zooplankton excretion rate, and Nz is the zooplankton to nitrogen conversion factor. All other terms defined as above. I do not differentiate between forms of nitrogen. The main factor determining the nitrogen uptake by diatoms in the surface mixed layer is the N concentration at which the diatom growth rate is one half the maximum (Hn). H n is taken to be equal to an 25 intermediate value of 1.0 mmol m , as estimated for diatoms (Jamart et al. 1977; Parsons et al. 1984). Upwelling of nitrogen: Wind-induced upwelling in the La Perouse region is one of two major sources of surface layer nitrogen. The other major source is from cross-shelf mixing of Juan de Fuca Strait outflow (Mackas et al. 1987; Crawford and Dewey 1989). For simplicity, I assume that upwelling is the main source of surface layer nitrogen. The daily amount of N added to the mixed layer is calculated, after Parsons (1988), by multiplying the ratio of the maximum upwelling rate (U) to the mixed layer depth, with N below the surface mixed layer (N3 0), and scaled to the Ekman transport rate pattern (Te). Freeland and Denman (1982) estimate that an Ekman transport rate of 7 m s"1 100 m"1 is equivalent to an upwelling rate of 0.5 m d for the Eddy region. Since the maximum weekly Ekman transport rate calculated during 1985-89 is 55.9 m"3 s"1 100 m" 1 ,1 use a maximum upwelling rate (U) of 4 m d . This estimated upwelling rate is about one third that observed off Oregon, and is consistent with reduced alongshore wind intensities and Coriolis force associated with northerly latitudes (Huyer 1983). The maximum rate of N input in the model occurred during the week of greatest positive Ekman transport (Te); N input during the remainder of the upwelling period is proportional to the maximum. During winter downwelling, N is added to the model mixed layer according to v(N0 - N), where v is a mixing rate proportional to the weekly averaged negative Ekman transport, NQ is the water column nitrogen concentration observed in winter (e.g., > 10 mmol m ; Thomas and Emery 1986), and N is the nitrogen concentration in the mixed layer. Nitrogen excretion by zooplankton: Mann and Lazier (1992) suggest that zooplankton excretion is an important source of nitrogen for phytoplankton in coastal upwelling regions. The amount of N excreted by model zooplankton is a function of their daily food ration, to a maximum of 20% (after Frost 1987). To convert zooplankton ration from carbon to nitrogen units I use a constant C:N ratio of 6:1. 26 Light limitation: Light limitation on diatom doubling rate is also modelled using a Michaelis-Menten function (see equation 1). In preliminary simulations I used a half-saturation constant for light (Hj) in equation 1 equal to 10% of the maximum surface P (after Parsons and Kessler 1986). The constant Hj results in an unexpectedly large and early phytoplankton bloom. This result suggests that Hj is not constant over the year but is relatively higher in winter than summer, and/or that there may be shifts in the dominance of phytoplankton (e.g., flagellates to diatoms; Mackas and Sefton 1982). To simulate this effect, I use a cosine function that varies H( from a minimum of 50 JAE m s in mid-summer to a maximum of 10 jiE m"2 s"1 in mid-winter (after Jamart et al. 1977). Mixed layer light (L) in equation 1 is calculated from observed photosynthetically active radiation (P), and from the estimated mixed layer depth (M), and average mixed layer extinction coefficient (X). From Parsons and Takahashi (1973), I use: P XM The extinction coefficient (X) in equation Id imposes the effects of self-shading and general water clarity on diatom growth. The coefficient is calculated by adding together a constant describing light absorption by seawater to the plankton concentration in the mixed layer (PL) (X = 0.1 + .006*PL). The range of extinction values generated by the function are similar to those observed in the Juan de Fuca Eddy during the upwelling season. Offshore export of diatoms: The second term in equation 1 represents diatom loss to offshore export. Transport of coastal surface layer phytoplankton to offshore areas is supported by recent studies using satellite imagery (e.g., Chavez et al. 1991). I assume that diatoms are exported out of the Eddy region during the upwelling season as a function of the pattern and intensity of positive Ekman transport (Te). Walsh (1981) indicated that about 50% of the annual primary production off coastal Peru is transported offshore. Thus as a starting point, I set the maximum daily export rate (Td) of diatom biomass to Equation 2a: L = * (! _ e(-xM)) 27 50%. Maximum daily transport occurs during the week of greatest positive Ekman transport, while diatom loss during the remainder of the upwelling season is proportional to the maximum transport. Diatom sinking: The third term in equation 1 represents the sinking of diatom biomass from the surface mixed layer of the Eddy. In the absence of vertical mixing large phytoplankton cells sink faster than small cells (Malone 1980). Nutrient-limited diatom populations also have a higher differential density (between the particle and the fluid) than exponentially growing ones, and thus the former sink faster (Kiorboe 1993). Thus, the sinking of model diatoms is a function of the upwelling rate and N, where Sd is the maximum sinking rate (1 m d ), Z is the depth of the mixed layer, N is nitrogen in the mixed layer, and H n is the N half-saturation constant (1 mmol m"3). High N associated with upwelling results in little or no sinking loss of diatoms. Diatom loss to predation: The forth term in equation 1 represents diatom biomass loss to ingestion by copepods and euphausiids, and is discussed below in sections 1.5.2 and 1.5.3, respectively. For simplicity, I assume that only 75% of the total diatom biomass is available to zooplankton grazers at any one time. This assumption recognizes that certain diatom genera are unpalatable because of the silica frustules, or that diatom quality is low because of seasonal nutrient limitation (e.g., Harrison et al. 1990). Table 1.1 summarizes the parameter values and important environmental constants discussed in the diatom biomass dynamics section. 28 Table 1.1. Summary of diatom parameters and environmental constants used in the model and discussed in the diatom biomass dynamics section (1.5.1) (m: metres). Parameter description Diatom biomass (Jan 1) Diatom export rate Diatom sinking rate Diatom respiration rate N half-saturation constant Nitrogen (Jan 1) Mixed layer depth (summer) Mixed layer depth (winter) Maximum upwelling rate Nitrogen below mixed layer Symbol D Td sd Rd H„ N0 M M U N30 Unit mg Chla m"3 % d"1 m d"1 % d"1 mmol m mmol m m m m d"1 mmol m Value 0.5 50.0 1.0 10.0 1.0 4.0 15 100 4.0 10.0 29 1.5.2 Copepod biomass dynamics Total copepod biomass on the southern BC shelf is seasonally variable ranging from about 0.5 g DW m"2 in winter to about 5.0 g DW m"2 in summer (Mackas 1992). Model copepods are represented by the genus Calanus which can, at times, comprise about 50-80% of the copepod biomass on the southern BC continental shelf during the upwelling season (see Figure 9 in Mackas 1992). In the model, I use a starting (winter) copepod concentration of 0.75 g DW m . The equation representing the processes effecting copepod biomass is: Equation 3: Ct+1 = Ct + — AC (GCICQ(D-P)) clleEcQ(C-P» IjQ (C-Pj)) JWd - TJeC At { (Hc+D-P) J [ (He + C-P) ) {(Hj + C-Pj)} = copepod growth - euphausiid grazing -juvenile herring grazing -export of copepods. Where C is the copepod biomass, Gc is the copepod growth efficiency, Ic is the copepod maximum ingestion rate, Q is the effect of temperature on ingestion, H c is the copepod prey half-saturation constant, P t is a prey threshold concentration, Ie is the maximum euphausiid ingestion rate, E c is the availability of copepods to euphausiids, H e is the euphausiid grazing half-saturation constant, E is euphausiid biomass, J is the juvenile herring biomass, L is the juvenile herring ingestion rate, H: is the herring prey half-saturation constant, P: is the herring prey threshold, Wd is the wet weight to dry weight conversion factor, and T c is the maximal rate of offshore transport of copepods.. Copepod growth: The first term on the right hand side of the copepod biomass equation represents growth from ingesting diatoms, where Gc is the gross growth efficiency, Ic is the maximum ingestion rate, Q is the effect of temperature on ingestion, H c is the prey half-saturation constant, and P t is a prey threshold constant. 30 The functional feeding response of copepods (and euphausiids) is represented using a Michaelis-Menten (hyperbolic) function. This type of ingestion function has been used by several authors to describe the results of laboratory feeding experiments between zooplankton and their prey, and by several authors in plankton modelling studies (e.g., Parsons and Kessler 1986, Frost 1987). The gross growth efficiency (Gc) term used in equation 3 is defined as the proportion of ingested prey that is converted instantaneously into somatic and reproductive tissue (Parsons et al. 1984). The conversion of ingested material is assumed to be maximally efficient at some finite ration. There is some laboratory evidence that growth efficiency declines with a further increase in ration, but I assume for simplicity, that it is constant. With decreasing rations, growth efficiency declines linearly and becomes zero at a ration which just maintains the animal. This maintenance ration covers the costs of basal and active metabolism, specific dynamic action, and excretion (Ware 1982). At rations below maintenance concentrations, growth efficiency becomes negative and the animal loses body weight. Figure 1.7 summarizes the generalized changes in gross growth efficiency with prey concentrations. Maximum gross growth efficiency (GGE) in copepods ranges from 30 to 50% (Conover 1978, Parsons et al. 1984; GLOBEC Report No. 2, 1993); I use a mid-range value of 40%. Copepod growth efficiency below the maximum value is determined by estimating saturating and maintenance prey concentrations from weight and temperature dependent equations derived for herbivorous zooplankton (Huntely and Boyd 1984). In using Huntley and Boyd's (1984) saturating and maintenance prey concentration equations, I assume that Calanus populations could be represented by an average body weight. This simplifying assumption is necessary to avoid modelling detailed population dynamics of generic copepods (and euphausiids), for which few data exist. This general approach maintains the same level of complexity in each model compartment from zooplankton to fishes (Silvert 1981). I use a copepod body weight equivalent to roughly 31 + GGE 0 Maxi murr GGE ...... A / i X Saturating / ration ^ ^ r ^ . . . . . _ . Maintenance _ _ . _ _ „ •""^ ration 1 1 1 I I ! 1 Low Arbitrary prey concentration (ugC) High Figure 1.7. Overview of the function used to describe the relationship between zooplankton gross growth efficiency (GGE) and changing prey concentration. 32 one-third of the observed maximum adult weight, which is the size where zooplankton maximize their production (Parsons et al. 1984). An adult C. marshallae weighs about 100 Hg C, and thus an average model copepod weighs 30 |xg C (corresponding to a stage 4 copepodite; Landry and Lorenzen 1989). A 30 |ig C model copepod, experiencing the range of temperatures observed in the Eddy region during 1985-89, would require about 35-70 [ig C L"1 to maintain itself, and about 130-250 |ig C L"1 to be fully saturated, and thus achieve maximal GGE and growth. A first order estimate of the maximum ingestion rate of copepods (Ic) is calculated from the equation, ingestion rate = maximum growth/GGE. A temperature-dependent food-saturated estimate of maximum growth (G'm a x ) for small zooplankton was derived by Huntley and Boyd (1984) as: G'm a x = 0.0542 e 0 1 1 T , where T is temperature (°C). Thus, a maximum growth rate of about 20% per day is estimated for maximum water temperatures observed in the Eddy region, and a maximum food-saturated ingestion rate of 50% body weight per day (BWD) is used in the model. This estimate falls within the 40-60% BWD range for "small" crustaceans cited in Parsons et al. (1984). The effect of water temperature on copepod (and euphausiid) ingestion is modelled assuming a Q1 0 of 2.0 (Parsons et al. 1984). Thus in the model, zooplankton ingestion doubles for every 10°C rise in water temperature. The concentration of prey that results in one-half of the maximum zooplankton ingestion rate (e.g., Hc) is set to 20% of the zooplankton saturating prey concentrations. This approach ensures that the functional feeding response maintains the same shape over the range of saturating concentrations, as the latter changes with water temperature (see above; after Jamart et al. 1977). There is evidence that zooplankton ingestion ceases at low prey concentrations (Parsons et al. 1984). Although the prey threshold concentration P t used in the zooplankton ingestion function is small (10 |ig C l"1), I found it to be particularly important when plankton feeding interactions are closely coupled during winter (See also 33 Evans and Parslow 1985; Frost 1987). Interestingly, Hofmann and Ambler (1988) noted that a feeding threshold for large phytoplankton (> 10 |im) was not required when detailed physical dynamics (e.g., advective and diffusive effects) were included in a plankton trophodynamics model. Copepod loss to predation: The second term in equation 3 represents euphausiid predation on copepods. Since euphausiids are known to feed primarily on smaller stages of copepods (i.e., nauplii or copepodites; Ohman 1984, Stuart and Pillar 1990), I assume that only a portion of the total copepod biomass is ingested by euphausiids at any given time (Ec). I arbitrarily set E c to a maximum of 40% of copepod biomass. E c is fully realized when copepod production is maximal, as determined by the saturating prey (diatom) function described above. Thus, increased copepod production resulted in a larger potential biomass of smaller stages susceptible to euphausiid predation. Copepods also experience morality from other invertebrate predators like chaetognaths, or gelatinous zooplankton like ctenophores (e.g., Ohman 1986). However, because concentrations of these invertebrate predators are relatively low during the upwelling season in the Eddy region (e.g., < 0.75 g DW m and < 0.5 g DW m , respectively; Mackas 1992), and because these predators are size selective, predation effects on larger copepods like Calanus are assumed negligible and ignored. Copepods like C. marshallae can make up a large proportion of the diets of juvenile fish (e.g., Pacific herring; Wailes 1936). Since one of the largest known biomasses of juvenile fishes on the southern BC continental shelf are juvenile herring (see section 1.5.5), I include their feeding impact on copepods (third term in equation 3). The estimate of copepod biomass eaten by juvenile herring is only first-order because growth and ingestion rates of the latter in the La Perouse region are relatively unknown. In the model, juvenile herring ingestion is primarily used to calibrate simulated autumn copepod biomass to the observed seasonal copepod biomass pattern (see Chapter 2). Copepod offshore transport: The last term in equation 3 represents the offshore 34 export of copepods via Ekman transport (Te). Since most copepod species and stages reside in the upper oceanic layer (< 50 m; Mackas 1992), I assume that model copepods are horizontally exported from the Eddy to the outer slope region as a function of the seasonal pattern and intensity of positive Ekman transport. Copepods are exported (Tc) from the surface waters of the Eddy region at a maximum rate of 13% d"1 (Mackas 1992). In addition, the maximum export rate occurs during the week of greatest Ekman transport (T). Table 1.2 summarizes parameter values discussed in the copepod biomass dynamics section. 35 Table 1.2. Summary of zooplankton parameters used in the model and discussed in the copepod biomass dynamics section (1.5.2), and the euphausiid biomass dynamics section (1.5.3) (DW: dry weight; |xg C: micrograms carbon; BWD: body weight per day). Parameter description Copepod biomass (Jan 1) Copepod body weight Maximum copepod growth efficiency Copepod maximum ingestion rate Copepod maximum export rate Proportion copepod biomass available to euphausiids Euphausiid biomass (Jan 1) Euphausiid body weight Maximum euphausiid growth efficiency Euphausiid maximum ingestion Euphausiid import rate Temperature effect on zooplankton grazing Zooplankton excretion rate Zooplankton prey threshold Symbol C --Gc Ic Tc E c E --Ge le Re Q xz Pt Unit g DW m"2 HgC % %BWD d"1 d"1 g DW m"2 HgC % %BWD g D W m " 2 d _ 1 — d"1 Hg C L"1 Value 0.75 30 40.0 50.0 0.13 0.40 2.0 1200 20.0 20.0 0.1 2.0 0.2 10.0 36 1.5.3 Euphausiid biomass dynamics Euphausiid biomass along the continental shelf off the west coast of Vancouver Island is primarily dominated by two species, Thysanoessa spinifera and Euphausia pacifica (Fulton et al. 1982). T. spinifera is generally more abundant than E. pacifica in nearshore areas of the continental shelf (Mackas 1992), a fact emphasized by T. spinifera's dominance in the hake diet during summer (Rexstad and Pikitch 1986; Tanasichuk et al. 1991). Peterson and Miller (1975) also indicate that populations of T. spinifera are about one order of magnitude higher than E. pacifica on the inner shelf off Oregon. In this modelling study then, Eddy euphausiid production dynamics should best be represented by T. spinifera. Unfortunately, the life history of T. spinifera has not been well studied, and thus I have supplemented data gaps with information from studies conducted on E. pacifica. Estimates of total euphausiid biomass on the southern BC shelf from hydroacoustic and Bongo net sampling range from about 1-2 g DW m"2 in winter to > 4 g DW m"2 in summer (see Chapter 2; Simard and Mackas 1989; Mackas 1992). I use a starting (winter) euphausiid biomass of 2.0 g DW m . The equation describing the processes effecting the model euphausiids is: LE Equation 4: E,.. = E+ — ' ' At A £ (GeIeQEc(D + C-P)\ _(IhQ(E-Pt)) c HhHE-P) (IsQSe)SWc + ReDc HWC -(IwQWe)WWc At { (He+D+C-P) ) "ft euphausiid growth - adult herring grazing - hake grazing - dogfish grazing + import Where E is the euphausiid biomass, Ge is the euphausiid growth efficiency, Ie is the maximum euphausiid ingestion rate, E c is the copepods availability to euphausiids, H e is the euphausiid prey half-saturation constant, D c is the dry weight to carbon conversion 37 efficiency, H is the adult herring concentration, Ih is the adult herring ingestion rate, P t is a prey threshold, H h is the herring prey half-saturation constant, H is the adult herring biomass, Wc is the wet weight to carbon conversion factor, I w is the hake ingestion rate, W is the hake biomass, We is the proportion of euphausiids in the hake diet, S is the dogfish biomass, I s is the dogfish ingestion rate, Se is the proportion of euphausiids in the dogfish diet, Re is the euphausiid import term, D c is the dry weight to carbon conversion factor. Euphausiid growth: The first term on the right hand side of the euphausiid biomass equation represents euphausiid growth, where Ge is the gross growth efficiency, Ie is the maximum ingestion rate, E c is the proportion of copepod biomass available to euphausiids, Q is the effect of water temperature on ingestion, H e is the half-saturation constant for diatoms (D) and copepods (C), and P t is the prey threshold concentration. See section 1.5.2 for a general discussion concerning Q, H e , and P t . The maximum euphausiid gross growth efficiency (Ge in equation 4) is difficult to ascertain, but Sushchenya (1970, in Parsons et al. 1984) indicates that GGE in E. pacifica ranges from 7% to 30%, (20% average), and Parsons et al. (1984) suggest that for most marine zooplankton GGE ranges 10-40%. I use an initial Ge of 20% in the model. Recall that copepod GGE varies in relation to temperature-dependent saturating and maintenance prey concentrations (see section 1.5.2). I use the same assumption and the Huntley and Boyd (1984) equations to estimate temporally variable euphausiid GGE. This is justified because Huntley and Boyd (1984) included data for E. pacifica in deriving their prey concentration equations. Again recall that the Huntley and Boyd (1984) equations are weight-dependent, and thus model euphausiids must be represented by an average body weight. The model euphausiids could best be represented by a 12-13 mm animal because this is the modal length of T. spinifera in March/Apri l (Fulton et al. 1982), and the average of the bimodal length-frequency distribution of euphausiids observed in June (Simard and Mackas 1989). A 12-13 mm euphausiid weighs about 1200 ng C using the length-weight 38 relationship derived by Ron Tanasichuk (DFO, Nanaimo) for T. spinifera collected in Barkley Sound: ln(mg WW) = -4.378+ 2.87 ln(length in mm). I assume that 6.75% of euphausiid WW is carbon. A first-order estimate of the maximum euphausiid ingestion rate (Ie in equation 4) is determined from Ingestion = maximum growth/GGE. Estimates of maximum growth rates for T. spinifera are rare but I derived one from Figure 8 in Fulton et al. (1982). T. spinifera were modally 10 mm on 15 March and 15 mm on 18 April; sampling was conducted at night off northern coastal BC. Using the above length-weight regression for T. spinifera, and %BWD = [ io( l o g W t 2 - l o g W t l V t i m e - l ] * 100 (Parsons et al. 1984), the growth rate is about 4% BWD. The estimated growth rate of 4% BWD is high for an adult euphausiid but reasonable considering that a T. spinifera is about 1.5 times as heavy as a similarly sized E. pacifica (Jamieson et al. 1990; i.e. weight differences imply T. spinifera has higher growth). In fact, a 12-13 mm E. pacifica has a growth rate of only about 2% BWD (Ross 1982; see also Table 2.3). A maximum model euphausiid growth rate of 4% BWD and GGE of 20% results in an ingestion rate of 20% BWD. This estimated ingestion rate is near or above the high end of the range observed for E. pacifica (7-20 % BWD; Lasker 1966; Parsons et al. 1967; Ross 1982; Ohman 1984), and for another coastal upwelling euphausiid, E. lucens ( 5 -15% BWD; Stuart 1986; Stuart and Pillar 1990). Euphausiid loss to predation: The next three terms in equation 4 represent the loss of euphausiids to predation from Pacific hake (W), adult Pacific herring (H), and spiny dogfish (S). I discuss the fish ingestion functions and parameter values for the above predators in sections 1.5.4, 1.5.5, and 1.5.6, respectively. Euphausiid import: Model euphausiids are not horizontally transported out of the Eddy region like copepods or diatoms because they undergo extensive diel vertical migrations that will tend to keep them in basins on the continental shelf (Simard and Mackas 1989; Pillar et al. 1992). Instead, the last term in equation 4 represents the import of euphausiids from surrounding oceanic regions due to upwelling. The rationale behind 39 euphausiid import is discussed in section 2.2.3. Table 1.2 summarizes parameter values discussed in the euphausiid biomass dynamics section. 1.5.4 Hake biomass dynamics The largest biomass of pelagic fish found on the continental shelf off southwestern Vancouver Island during summer is the Pacific hake. Hake made up about 65% of the total pelagic fish biomass during the 1980s (Ware and McFarlane 1994). The authors used swept-volume data from trawl surveys to estimate an average 1980s biomass of 236,000 tonnes (236 kt). Roughly 90% of this total hake biomass occurs in the basins of the Eddy region (100-200 m; see Figure 8 in Ware and McFarlane 1994). I discuss the timing of arrival of the 210 kt hake biomass to the Eddy region below. The equation describing model hake biomass dynamics is: Equation 5: Wt+1 = Wt + — 11 ' At ™ = (JWGWQ)W - (SJSQ)S - Fw +/~ Mw = growth - dogfish grazing - fishery ± migration Where W is the hake biomass, Iw is the hake ingestion rate, G w is the hake growth efficiency, S is the dogfish biomass, Is is the dogfish ingestion rate, Sw is the proportion of hake in the dogfish diet, F w is the hake fishery, and M w is the immigration or emigration of hake. Hake growth: The first term on the right hand side of the hake biomass equation represents growth where I w is the hake ingestion function, Gw is the gross growth efficiency, and Q is the effect of temperature on hake ingestion (Q1 0 = 2.0). The hake ingestion function (Iw) is equal to the proportion of prey in the hake diet 40 multiplied by the hake ingestion rate or ration. On the southern BC shelf, euphausiids and herring make up roughly 70% and 24% respectively, of the prey in the hake diet during August (1985-89 average volume; Tanasichuk et al. 1991). Bailey et al. (1982) also indicate that euphausiids dominate the summer diets of migratory hake in the northern Coastal Upwelling Domain. Tanasichuk et al. (1991) hypothesized that euphausiid biomass on the southern BC shelf determines the proportion of euphausiids and herring in the hake diet (see also Livingston 1983). The authors contend that aggregations of euphausiids in upwelling regions attract hake, and other planktivores like herring, which then become incidental prey of the hake. Data in Tanasichuk et al. (1991) indicate that the fraction of euphausiids (herring) in the hake diet is highest (lowest) in June/July (90% versus 4%) before decreasing (increasing) to about 60% (14%) in September/October. This pattern corresponds to seasonal highs and lows in euphausiid biomass (See Chapter 2). Thus, the fraction of euphausiids in the model hake diet is a function of euphausiid biomass. For simplicity, if model euphausiid biomass is > 60 kt (75% of maximum estimated euphausiid biomass) then the proportion of euphausiids in the hake diet is constant at 95%. As euphausiid biomass declines the fraction in the hake diet decreases to 60% at 24 kt of euphausiids. The herring portion of the hake ration is equal to one minus the euphausiid portion. In addition, 60% of the herring ration consists of adults, while about 40% consists of juvenile herring (Shaw et al. 1987, 1989a,b). Hake ration is positively correlated to estimated euphausiid biomass in the Eddy region (r = 0.91; Figure 1.8). The highest hake ration observed in June (4% BWD) corresponds with the highest estimated euphausiid biomass (80 kt), while the lowest hake ration in July matches the lowest seasonal euphausiid biomass estimate. For simplicity, I assume that hake ration remains constant at 4.0% BWD for euphausiid biomass > 80 kt. Model hake ration also declines linearly with euphausiid biomass ration such that it is about 1.25% BWD at a euphausiid biomass of 40 kt. Note as discussed above, the hake 41 a « 4.0 C o •E 3.0 o a 2 W 2.0 (0 sz a 3 a> 1.0 <u j * cs X • E u p h a u s i i d •*" Herring — • --» ^ m ^ ^ ^ ^ m ^ A l I l — I ^ ^ ^m m ~x ^ ^ » A I I I I I I ^ h 0.5 0.4 0.3 0.2 0.1 a c o "J 1 -o a en c JZ • (0 X 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 Euphausiid biomass (gDW/m2) Figure 1.8. Proportion of euphausiids and herring in the hake ration plotted against estimated euphausiid biomass for June to September (See Chapter 2). 42 ingestion of herring is inversely related to the estimated euphausiid biomass (r : 0.51; Figure 1.8). The average 1985-90 August hake ration of 1.6% BWD estimated by Tanasichuk et al. (1991) is about twice as high as that estimated for 45-55 cm hake by Rexstad and Pikitch (1986) using equations from Francis (1983), but is consistent with estimates for other gadoids (e.g., Table 10 in Francis 1983). Empirical data describing the saturating and maintenance rations of marine fish are generally not available in the literature. I assume that fish GGEs are constant over time. In addition, maximum GGE for hake is 10%, which is one-half that used for herring. This is consistent with evidence that GGE declines as organisms increase in body weight (Parsons et al. 1984). Hake growth rate is calculated by multiplying together ingestion rate and GGE. The average observed hake ration in August is 1.6% BWD (Tanasichuk et al. 1991), and thus hake growth is estimated at 0.16% BWD. Accurate empirical estimates of hake growth are difficult to obtain because of the highly migratory nature of the stock (Smith et al. 1990; see also Section 2.3.1). Hake loss to predation: The second term in the hake biomass equation represents the loss of hake to dogfish predation, where Sw represents the proportion of hake in the dogfish diet. The latter is a function of the proportion of euphausiids in the dogfish diet (see section 1.5.6). Hake fishery: The third term in equation 5 describes the biomass of hake removed by the seasonal offshore joint-venture fishery (Fw). I assume that the hake fishery operates at a constant rate (420 t/day) from 1 June to 31 October (150 days). The total biomass of hake removed is equivalent to the 1985-89 average catch of about 63 kt (Ware and McFarlane 1994). Hake migration: The final term in the hake biomass equation represents the immigration of hake to, or emigration from, the Eddy region (Mw in equation 5). The coastal Pacific hake stock spawns off southern California in the first quarter of the year, and then migrates to feeding grounds in the northern Coastal Upwelling Domain; larger, 43 older fish migrate further north (Bailey et al. 1982). It is presently unclear as to the exact factors influencing the timing of arrival of the migratory hake stock to coastal BC waters, but it is likely linked to water temperature and/or currents (Bailey et al. 1982; Beamish and McFarlane 1985; Smith et al. 1990). A first-order estimate of the arrival time of migratory hake to the southern BC shelf comes from data in the La Perouse 1990 Annual Report. Migratory hake appear in gill net and balloon trawl catches, conducted during 1987-90 in Barkley Sound, from mid May until late July. I assume model hake immigrate into the Eddy region at a constant rate over 30 days, starting 1 June. While migratory hake occupy the northern feeding grounds they exhibit a seasonal inshore/offshore migration (e.g., Bailey et al. 1982). Direct evidence of seasonal cross-shelf movement of hake in coastal BC comes from tracking the position of the hake fishing fleet (see Ware and McFarlane 1994). The hake fleet is located in nearshore regions of the continental shelf (e.g., Eddy region) in early summer, and then it moves offshore in autumn. Additional evidence supporting the inshore/offshore seasonal hake migration comes from noting that hake primarily consume T. spinifera in spring/summer, and E. pacifica in autumn (Ron Tanasichuk, DFO Nanaimo, unpublished data). The shift in euphausiid species composition in the hake diet can be related to the fact that T.spinifera is most abundant at nearshore locations, while E. pacifica is more abundant at the shelf/slope break (Fulton et al. 1982). Tanasichuk et al. (1991) linked cross-shelf hake migration to euphausiid biomass on the southern BC shelf. The authors indicate that reduced hake rations (see above) in autumn correspond to reduced nearshore biomass of T. spinifera. Offshore areas are thought to be more important in late summer or autumn after euphausiid biomass on the shelf has been depleted by hake predation. I assume that hake emigration from the Eddy to offshore regions during summer/autumn is inversely related to euphausiid biomass in the Eddy region. The maximum model hake emigration rate (5.0% of the biomass/day) is estimated from simulations resulting in 100 kt of hake remaining in the Eddy region by 44 the end of August; this hake biomass corresponds to hydroacoustic estimates made at that time (Ware and McFarlane 1994). Model hake do not emigrate from the Eddy region if euphausiid biomass is arbitrarily > one-half of the maximum euphausiid biomass observed in June. Pacific hake undergo a return migration from northern feeding regions to southern spawning grounds in late autumn. This southward migration is thought to be linked to the transition from summer to winter oceanic currents (Bailey et al. 1982). Hake remaining in the Eddy region after the autumn transition are removed at a rate of 50% of the biomass per day. 1.5.5 Pacific Herring biomass dynamics Pacific herring are the most abundant resident fish species found on the continental shelf off southwestern Vancouver Island. In the model I consider two groups of herring: adults (>3+ yr) and juveniles (0, 1+, and 2+ yr). During the 1980s, adult herring constituted about 25% of the total fish biomass during summer (55 kt; Ware and McFarlane 1994). Juvenile herring are included in the model because they represent a large biomass of small fishes that feed on copepods. Juvenile herring biomass is assumed to be equivalent to the adult herring biomass (55 kt). The two equations used to describe processes effecting adult and juvenile herring biomass are the same except that adults feed on euphausiids, while juveniles feed on copepods (Wailes 1936). The general adult herring biomass equation is: Equation 6: HHl = Ht + AH _ (GhIhQ(E-PJ\ H - (IwQ(l'We)WhW) - (IsQ(l-Se)ShS) - (I0QOhO) + / - Mh At { (Hh+E-PH) J = growth - hake grazing - dogfish grazing - salmon grazing + or - migration 45 The general juvenile herring biomass equation is: Equation 7: Jt+1 = Jt + — ±L = \_L££L Jl j - (IwQ(l-We)WjW) - (JsQ(l-Se)SjS) - (I0QOjO ) + / - Mj At { (HJ + C-PJ) , = herring growth - hake grazing - dogfish grazing - salmon grazing ± herring migration Where H is adult herring biomass, J is juvenile herring biomass, Ih and L are herring ingestion rates, G h and G; are herring growth efficiencies, H h and H: are herring prey half-saturation constants, P h and P: are herring prey threshold concentrations, Q is the temperature effect on herring ingestion, M: and M h are herring migration rates. All other terms as defined above. Herring growth: Adult herring feed entirely on euphausiids, based on stomach-contents data I collected while on the "MV Howe Bay" in 1990. Wailes (1936) and Shaw et al. (1987) also note that adult herring off southern Vancouver Island feed primarily on euphausiids during the summer. For simplicity, I assume that juvenile herring feed entirely on copepods. The maximum ingestion rate of adult herring is determined from: ingestion = observed growth/GGE. Growth of age 4 herring from 1 March to 1 October averaged over 1986-88 was 0.32% BWD (D. Ware, DFO, Nanaimo, unpublished data). Assuming a constant maximum GGE of 20% for adult herring the maximum ingestion is calculated at 1.6% BWD. Simulations indicate that juvenile herring need to ingest about 3% BWD to generate realistic copepod biomass patterns in autumn (see Chapter 2). The simulated juvenile herring ingestion rate is within the range (2-5% BWD) observed for coastal BC juvenile herring (Houde and Berkely 1981). Prey threshold values for the juvenile and adult herring ingestion functions are set equal to winter concentrations of copepods and euphausiids, respectively. In addition, 46 herring prey half-saturation constants are estimated to be 1.5 to 3 times the prey threshold values. Herring loss to predation: The next three terms of equations 6 and 7 represent loss of adult and juvenile herring to predation by hake, dogfish, and salmon, respectively. The proportion of adult or juvenile herring in the hake, dogfish, and salmon diet is determined from data collected on the southern BC shelf during 1985-87 (Shaw et al. 1987, 1989a,b). The proportion of herring in the diets of hake and dogfish is assumed to be a function of euphausiid biomass, where We and Se are the proportion of euphausiids in the hake (H) and dogfish (S) diet. On average about 60% and 40% of the herring portion of the hake diet consists of adult and juvenile herring, respectively. Chinook salmon on the southern BC shelf in summer feed predominantly (> 90%) on juvenile and adult herring (personal observation; Shaw et al. 1989a). The Chinook salmon herring ration consists of 80% adults and 20% juveniles (from data in Shaw et al. 1989a). Herring migration: The last terms in equations 6 and 7 represent herring migration. I assume that adult herring immigrate to the Eddy region from nearshore spawning areas (e.g., Barkley Sound) at a constant rate over 30 days starting 1 April (1833 t /day). To represent the migration to nearshore spawning areas, adult herring emigrate out of the Eddy region in December at a rate equal to the immigration rate. I assume that 50% of the 55 kt juvenile herring biomass consists of 1+ and 2+ fish, while the remainder are young-of-the-year (YOY). Juvenile herring remain in the Eddy region year-round, while YOY migrate to the Eddy region from nearshore nursery areas in mid-August (Ware and McFarlane 1994). At the end of December, 50% of the total juvenile herring biomass emigrate to nearshore spawning areas with the adults. 1.5.6 Biomass dynamics of spiny dogfish and chinook salmon Dogfish: The top-level predator included in the model is the spiny dogfish. Estimates of dogfish biomass along the west coast of Vancouver Island are difficult to 47 make because few fishery data exist, and the animals exhibit wide ranging migrations. Ware and McFarlane (1994) however, produced a first-order estimate of dogfish biomass on the southern BC continental shelf using swept-volume estimates from mid-water trawls surveys conducted in August. I assume that 50% of the 38 kt estimated dogfish biomass occurs in the Eddy region during summer. Ketchen (1986) noted that dogfish tend to migrate roughly seasonally along the coasts of BC and Washington. Thus, I assume that dogfish move into and out of the Eddy region in relation to the seasonal cycle in water temperature. Maximum model dogfish biomass (20 kt) occurs in August, while minimum concentrations (10 kt) occur in January. Dogfish ingestion is calculated by multiplying the ingestion rate by the proportion of prey in the diet. Maximum dogfish ration is taken from Jones and Geen (2% BWD; 1977). This ration is similar to that estimated for dogfish in August around La Perouse Bank (2.6% BWD; Tanasichuk et al. 1991). Together, euphausiids, herring, and hake constitute about 90% of the total dogfish diet in August (Tanasichuk et al. 1991). As for hake, the proportion of euphausiids in the dogfish diet is linearly and positively related to euphausiid biomass. The dogfish diet consists solely of euphausiids when Eddy euphausiid biomass is > 80 kt (June euphausiid maximum). The proportion of herring and hake in the dogfish diet is calculated as one minus the euphausiid proportion. In August, the dogfish diet consists of about 60% euphausiids, 16% hake, 5% adult herring, and 10% juvenile herring (Shaw et al. 1989a,b; Tanasichuk et al. 1991). Annual dogfish production is not calculated because of low growth rates (Ketchen 1986), and uncertainties in biomass estimates. Chinook salmon: Chinook salmon constitute about 3% of the biomass of pelagic fishes found on the southern BC continental shelf during summer (11 kt; Ware and McFarlane 1994). Chinook salmon are included in the model because they have a large predatory impact on juvenile and adult herring. In fact, > 90% of the diet of Chinook salmon consists of herring (Shaw et al. 1987). In the model, the 11 kt of Chinook salmon i 48 feed exclusively on adult and juvenile herring year-round. The Chinook salmon ingestion rate is set at 0.5% BWD. Production of salmon is not calculated because of the relatively small biomass. Table 1.3 summarizes parameter values discussed in the hake, herring, and dogfish/salmon biomass dynamics sections. 49 Table 1.3. Summary of fish parameters used in the model and discussed in the hake, herring, dogfish, and salmon biomass dynamics sections (1.5.4, 1.5.5, and 1.5.6, respectively) (kt: thousands of tonnes; BWD: body weight per day; DW: dry weight). Parameter description Hake biomass Hake maximum ingestion rate Hake growth efficiency Adult herring biomass (April 1) A. herring maximum ingestion A. herring growth efficiency A. herring prey half-saturation A. herring prey threshold Juvenile herring biomass (Jan 1) J. herring biomass (Aug 1) J. herring maximum ingestion J. herring growth efficiency J. herring prey half-saturation J. herring prey threshold Dogfish biomass (Jan 1) Dogfish ingestion rate Salmon biomass (Jan 1) Symbol W Iw Gw H h Gh Hh Ph J J Ji Gj Hj P i S Is o Unit kt %BWD % kt %BWD % g DW m"2 g DW m'2 kt kt %BWD % g DW m'2 g DW m"2 kt %BWD kt Value 210 4.0 10.0 55 1.6 20.0 3.0 2.0 28.0 28.0 3.0 20.0 2.3 0.75 20.0 2.0 11.0 50 CHAPTER 2: CALIBRATION, CORROBORATION, AND SENSITIVITY OF THE JUAN DE FUCA EDDY TROPHODYNAMICS MODEL 2.1 Introduction In Chapter 1 of my dissertation I discussed the structure and assumptions of the Juan de Fuca Eddy trophodynamics model. In the first section of this Chapter, I will outline the protocol for calibrating the model output to empirical data collected for the La Perouse region. In section 2.3,1 corroborate output from the standard run with observations of production properties made in other regions of the Domain, and in other coastal upwelling systems. Finally in section 2.4, I evaluate the sensitivity of the model to variability in abiotic and biotic parameters, and to assumptions in major processes like hake migration. 2.2 Model calibration Model output is primarily calibrated to the observed seasonal patterns in zooplankton biomass (see sections 2.2.2 and 2.2.3). Model output is secondarily calibrated to data describing the proportions of prey in fish diets, and to fish ingestion and growth rates. Simulated output is not expected to calibrate exactly to empirical data because I use average biological parameter values, and because there is temporal variability in biological sampling, and in the environmental forcing functions. Refer to section 2.4 for the statistical criteria I use to determine how one simulation is better calibrated than another simulation to the empirical data. I now discuss the calibrated (standard run) 1985-89 simulated output for plankton and fish. 2.2.1 Diatom biomass pattern Simulated diatoms are not calibrated to a specific empirical seasonal biomass 51 pattern because of a lack of data for the Juan de Fuca Eddy region during 1985-89. However, simulated diatom biomass is compared to individual observations of chlorophyll a concentrations in the La Perouse region, and to the general seasonal chlorophyll a pattern observed in the coastal upwelling region off the Washington coast. Simulated diatom biomass increases from a winter minimum of 0.5 mg Chla m to about 5 mg Chla m" in mid April because of increased average light in the surface mixed layer and low zooplankton grazing (Fig 2.1). Observations off the Washington coast suggest that a spring bloom of this magnitude and at this time is reasonable (e.g., Figure 1.24 in Landry and Lorenzen 1989). After the spring bloom decline due to nutrient limitation, simulated diatom biomass increases rapidly in May/June in response to the initiation of upwelling. Throughout the summer, several simulated diatom blooms result in consistently and relatively high chlorophyll concentrations (5-15 mg Chla m ; Figure 2.1). Chlorophyll a concentrations measured in the La Perouse region similarly range from 5 to 20 mg Chla n T 3 (Denman et al. 1981; Hill et al. 1982; Mackas 1992). 2.2.2 Copepod biomass pattern An extensive grid of about 15 stations has been used by DFO since 1985 to sample zooplankton around the continental shelf off southwestern Vancouver Island. The stations are visited opportunistically by ships from IOS and PBS, and are sampled using vertical or oblique bongo net hauls (0.23 mm mesh). These hauls integrate zooplankton biomass from near bottom to surface (see La Perouse Annual Reports 1986-90; Mackas 1992). I derive an empirical seasonal pattern of copepod biomass from data collected at three stations over deep water basins of the Eddy (C1,C2,C3), and an additional basin station (B8) located northeast of La Perouse Bank (Figure 2.2). These data are kindly supplied by D. Mackas (IOS, Sidney). Combining B8 with the three Eddy stations is justified because all 4 stations have similar zooplankton species composition (See Figure 4 in Mackas 1992). To develop a seasonal biomass pattern for copepods, I pooled the 52 Month Figure 2.1. Simulated weekly averaged diatom biomass pattern for the period 1985-89. 53 _s^ Figure 2.2. Location of sampling stations used to estimate the seasonal zooplankton biomass pattern for the Eddy region (See Mackas 1992). 54 available samples collected at the 4 stations for each month and calculate arithmetic averages; this protocol is used because not all stations are sampled in each month. Data for Calanus marshallae and Pseudocalanus spp are combined to describe seasonal changes in copepod biomass; these two species represent about 60-80% of the total copepod biomass (Mackas 1992). The copepod simulation captured the observed spring and early summer increase in copepod biomass as a direct result of growth from increased diatom production (see Figure 2.3). The general decline in copepod biomass during the summer is produced by the effects of offshore transport, as hypothesized by Mackas (1992). Initially however, simulated copepod biomass in late summer remained substantially higher than observed data even with offshore transport. In nature, this period corresponds to the arrival of a large biomass of juvenile herring to the La Perouse Bank region (Ware and McFarlane 1994). Since juvenile herring feed primarily on copepods (Wailes 1936), I included this process in the model. Simulated copepods more closely calibrate to the empirical data when juvenile herring feeding is included (Figure 2.3). Refer to the sensitivity analyses (Section 2.4) for an evaluation of other factors influencing simulated copepod biomass. 2.2.3 Euphausiid biomass pattern The observed seasonal changes in euphausiid biomass are shown, for stations C1,C2,C3,B8 of the Eddy region by month during 1985-89, in Figure 2.4. Biomasses of both T. spinifera and E. paciflca are relatively low in early summer (May to July) and increase to maximum concentrations in late summer to autumn (August to October; see also Mackas 1992). Several independent lines of evidence suggest that the empirical patterns in Figure 2.4 may not be fully capturing the seasonal dynamics in euphausiid biomass. Early summer euphausiid biomass may have been underestimated because of: i) known biases associated with sampling euphausiids during the day, ii) data describing Pacific hake ingestion, and iii) biomass from expected euphausiid population size structure during 55 ^ «* ^ ^ ^ ^ ^ ^ • o* ^ «* Month Figure 2.3. Simulated weekly averaged copepod biomass pattern calibrated to the 1985-89 empirical data. Boxes represent plus or minus one standard error around the mean sampling time and observed concentration. 56 25 20 Q 15 10 5.0 4.0 CM g 3.0 Q O>2.0 1.0 0.0 T. spinifera 8(2) 14(6) 6(0) 4(3) 8(6) 3(3) • 2(2) •f—m ^ <.*> ** ^ ^ /> v»s ^ #* o* ±* < / Month E. pacifica 7(D • 4(0) • : 12(9) • * •ytt 14(6) • 5(4) -5l^§-^""~~"\ 4(2) >»v « # ^ ^x ^< ^ _ K ^ ** v ^ ^ o* *«4 «* Month Figure 2.4. The observed seasonal biomass patterns for euphausiids in the Eddy region during 1985-89. The total number of samples collected (and daylight samples) are indicated. A * represents collections 1 h after sunrise and 1 h before sunset (daytime), while small boxes are collections 1 h after sunset and 1 h before sunrise (nighttime). 57 spring. The remainder of this section presents an overview that synthesizes available data and highlights potential underestimates of euphausiid biomass in spring and early summer. Night I day catch differences: Evidence that euphausiid biomass in the Eddy region is underestimated comes from the fact that the majority of early summer samples were collected during daylight, while late summer and autumn samples were collected primarily during the night (Figure 2.4). The time of collection strongly influences euphausiid biomass estimates because of the animal's vertical diel migratory behaviour. Generally, larger stages of most euphausiid species reside in relatively deep (> 100 m) water during the day. The daytime depth is influenced by light levels, water temperature, oxygen concentration, and other factors (e.g., Anderson and Nival 1991). Euphausiids are found primarily in near surface waters at night, while the animals vertically migrate to/from day and night depths during crepuscular periods. Given the vertical migratory behaviour of euphausiids one would expect night collections to contain more biomass than day collections. In fact, perusal of the literature indicates that night samples can maximally yield, on average, between 15 and 28 times more euphausiid biomass than corresponding day samples (Table 2.1). The variability in night/day catch ratios among studies is likely due to inherent patchiness in abundance, seasonal and interannual effects, and different species and sizes of euphausiids. It is generally accepted that higher night-time biomass collections are due to increased availability of euphausiids and/or to less visual avoidance of the sampling gear. Everson and Bone (1986), for example, noted that E. superba's ability to detect an oncoming net was hampered by darkness. To evaluate night/day catch effects on euphausiid biomass estimates, I analyzed zooplankton samples collected from Eddy stations (C1,C2,C3,B8) and shelf stations (B2,B3,B4,B5,B6,B7) of the La Perouse region. The data are sorted into samples collected during the night and those collected during the day. Night is defined as the period 1 h after sunset to 1 h before sunrise, while day is 1 h after sunrise to 1 h before sunset; 58 Table 2.1. Observed maximum and average differences in night versus day catches of euphausiids using various sampling gear. In each case night biomass estimates of euphausiids exceed day estimates by the amount indicated. Gear abbreviations are: BN: bongo net; PN: plankton net; MN: MOCHNESS; PL: plummet net; BI; BIONESS. Maximum euphausiid lengths: Euphausia pacifica 25 mm; Nematoscelis megalops 25 mm; Thysanoessa longipes 30 mm; T. gregaria 30 mm; Meganyctiphanes norvegica 31 mm; T. spinifera 34 mm. (N: night, D: day, X: times). Species E. pacifica E. pacifica E. pacifica (St. of Georgia) E. pacifica (WCVI) E. pacifica (October) E. pacifica (February) N. megalops T. longipes M. norvegica T. gregaria MEAN (maximum values) MEAN (average values) T. spinifera Gear BN PN BN BN BN PL PL MN BN BI PN BN Catch ratio N 10X > D N 20X > D N 20X > D N 20X > D N 10X > D N 31X > D (av: 11X) N 64X > D (av: 30X) N 63X > D (av: 15X) N 6X > D N 14X > D (av: IX) N 50X > D (av: 23X) N 28X > D N 15X > D N 20X > D Reference Youngbluth (1976) Brinton and Townsend (1981) Fulton et al. (1982) Hovekamp (1989) Wiebe et al. (1982) Fulton et al. (1982) Cochrane et al. (1991) Brinton (1967) This study 59 sunrise and sunset are determined from the Nautical Atlas for 50°N. Data falling within the two crepuscular periods are excluded from all analyses. This protocol prevents potential bias of overestimating biomass as a result of vertically migrating euphausiids (e.g., Alton and Blackburn 1972). Mean and standard deviations are calculated for night and day catches for the Eddy and shelf regions. The results indicate that T. spinifera biomass is about 20-30 times higher when sampled at night than during the day (Table 2.2). It should be noted that inclusion of crepuscular data reduces the night/day catch ratios by 30-40%; this results from higher daytime biomass estimates and indicates that sampling around sunrise or sunset captures vertically migrating euphausiids. The large standard deviations of night and day catches (Table 2.2) indicate that other factors like patchiness, interannual, seasonal effects, and body size greatly influence biomass estimates during day or night. A comparison of coefficients of variation in night versus day catches of euphausiids collected in the Eddy region during the same year and season, (Fulton et al. 1982), with the 1985-89 data indicates similarly high mean variability (188% versus 119% for day samples, and 228% versus 176% for night samples, respectively). This suggests that euphausiid patchiness might be the dominant factor contributing to observed sampling variability. Biomass variability due to patchiness cannot be 'corrected for' but only reduced by increased sampling (e.g., Watkins et al. 1990). This implies that correcting for night/day catch differences would increase the precision of euphausiid biomass estimates but not their accuracy (Sokal and Rohlf 1981). To further estimate the effects of seasonal bias on night/day catches, I examined euphausiid samples collected from the Eddy stations during August only and October/November only. This analysis indicates that night time estimates of T. spinifera biomass are 10 times and 40 times > daytime estimates, respectively. Hovekamp (1989) also noted a linear increase in night/day catch ratios for E. pacifica from August to February; no explanation is offered for the trend. Caution must be used in interpreting 60 Table 2.2. Estimates of zooplankton biomass (mg DW m"3) collected by Bongo nets in the Eddy and shelf regions during night and day (AV: average; SD: standard deviation; N: number of samples; N / D : night/day ratio). T. spinifera E. pacifica Copepods EDDY STATIONS N D N/D SHELF STATIONS N D N/D AV SD N AV SD N AV SD N 35.3 48.0 19 3.8 8.3 19 4.6 5.4 22 1.7 1.7 22 1.5 2.1 23 15.0 18.7 41 21 2.5 0.3 18.4 57.0 17 2.1 4.3 9 14.0 16.1 33 0.6 0.6 9 0.9 2.1 8 20.6 24.7 35.0 30 2.4 0.7 61 the results from my analysis because of small N, and because it is impossible to separate the effects of interannual variability. The few Eddy region data that do exist in any one month and year are greatly biased with respect to time of collection. For example, the majority of 1985 samples are collected during the day, whereas in 1989 they are mainly collected at night. A rigorous evaluation of interannual differences in night/day catch differences is thus not possible. A simple test of the validity of the derived first-order euphausiid night/day catch ratio involves comparing it to copepod data. I expected a priori that copepod samples should contain similar biomass estimates irrespective of time of collection. This is because most copepod stages and species reside within the upper 50 m of the water column during day and night, and thus are equally available to the sampling gear (Mackas 1992). As expected there are marginal differences between day and night biomass estimates (Table 2.2). Although the copepod night versus day biomass means are not statistically different it is interesting to note that day catches are about 1.5 to 3 times > night catches. Brinton and Townsend (1981) also noted increased catches of non-vertically migrating larval E. pacifica during the day. Are smaller zooplankton more aggregated in near surface waters during the day when large zooplankton biomass and subsequent predation risk is lower? A comparison of E. pacifica's and T. spinif era's catches indicates that the latter species has a 10 times higher night/day ratio (Table 2.2). The difference in night/day catch ratios between the two euphausiid species is consistent with what has been hypothesized about T. spinif era's feeding behaviour. During the day T. spinifera may feed on detrital matter on, or very near, the sea floor (D. Mackas person, comm; see also Cochrane et al. 1991). Thus, T. spinifera are less available to 'pelagic' samplers (Bongo nets) during the day than at night when the animals feed in surface waters. Apparently, the omnivorous E. pacifica maintains itself in the water column and thus is more susceptible to sampling with Bongo nets day or night. A combination of availability and net avoidance are likely responsible for the different euphausiid day/night catch ratios. 62 Hake ingestion: Additional evidence that euphausiid biomass may be underestimated in the Eddy in early summer is derived indirectly from consumption of euphausiids by dominant pelagic fish like Pacific hake. I assume that the pelagic fish are sampling (eating) euphausiids in proportion to the latter's availability. This assumption is not unreasonable for hake, herring and/or dogfish given that euphausiids make up large fractions of their diets (> 60-70%; Tanasichuk et al. 1991). The ingestion rate of hake and the fraction of euphausiids in the hake diet both suggest that there is a relatively high biomass of euphausiids in early summer, and substantially fewer euphausiids in late summer and early autumn (see below). This trend is opposite that observed in data derived from sampling with Bongo nets. The discrepancy between euphausiid biomass patterns predicted from fish consumption estimates and from net sampling may be related to the fact that each "method" is sampling a different portion of the euphausiid population. Tanasichuk et al. (1991) for example, noted that the majority of euphausiids making up fish diets were larger adults (> 20 mm). Large euphausiids may not be effectively sampled by Bongo nets because the animals avoid the small net mouth opening or avoid the vibration from the net bridle, among other factors (Wiebe et al. 1982). Data in Brinton and Townsend (1981) indicate that oblique tows of Bongo nets did not capture T. spinifera > 21 mm; the net most efficiently sampled individuals < 17 mm. Other studies have also noted differences in euphausiid population size structure when comparing net catches and euphausiid predator diets. Daly and Macaulay (1988) captured "small" E. superba (< 25 mm) with a plummet net, but generally "missed" the majority of biomass of large euphausiids (> 25 mm). The authors also noted that seal and sea-bird stomachs consisted primarily of larger euphausiids (35-55 mm); a situation perhaps not unlike the Eddy region in early summer (See also Cochrane et al. 1990). Euphausiid population size structure: Final support for the idea that euphausiids, particularly larger adults, are under sampled during early summer is based on expected 63 population size structure. Euphausiid populations in the Eddy region in early summer consist primarily of relatively older, larger adults (> 1 y and 20 mm) and a mode of younger, smaller individuals (< 1 y and 10 mm; e.g., Simard and Mackas 1989). By late summer or early autumn, the euphausiid population consists primarily of smaller, younger individuals, because larger, older stages are eaten or die. Day (1971) similarly observed on the southern BC shelf that lower euphausiid biomass in November, compared to higher biomass in May, was due to the absence of larger individual T. spinifera (Figure 2.5), and E. pacifica (not shown). The empirical euphausiid biomass estimates in late spring/early summer are thus consistent with what is expected from net avoidance bias and euphausiid population size structure. That is, net derived euphausiid biomass estimates are low in early summer because juvenile biomass is low, while late summer/autumn biomass estimates increase because more susceptible stages of euphausiids are available to the Bongo net. Cochrane et al. (1991) also noted for the Scotian shelf that "...a consistent pattern characterized by a greater proportion of smaller, more juvenile euphausiids in the autumn season. The comparatively low June catches may be partly due to more efficient net avoidance by larger adult euphausiids." Adjusted euphausiid seasonal biomass pattern: To generate a euphausiid seasonal biomass curve that is more consistent with the above evidence, I applied the first-order night/day catch ratio (20 times) to all day samples of T. spinifera and (2.5 times) to all day samples of E. pacifica. All crepuscular and night-time data were left unadjusted. This approach assumes that night-time net sampling is 100% efficient at catching euphausiids; undoubtedly however there is even some net avoidance at night. The adjusted seasonal biomass patterns (Figure 2.6) differ from the unadjusted biomass patterns (Figure 2.4) mainly in that the estimated euphausiid biomass in late spring/early summer (April-July) is substantially higher. This results from adjusting a large number of daytime samples collected during this period. 64 25 20 «- 15 C <D U 0) 0- 10 \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ a 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Length (mm) Figure 2.5. Size frequency of T. spinifera caught using an Isaac-Kidd midwater trawl at night off southwestern Vancouver Island and northern Washington (from Day 1971; modal lengths are 21 mm in May and 14-15 mm in November). 25 20 E Q O) 15 10 T. spinifera 65 8(2) X 14(4) • 6(0) • • 3(3) X X i 8(5) X y 2(2) X / * \ i 5(4) X • T • • ¥ X • • i B 4(3) • X * x 1 x ^ <? +* ^ +* #« *» ^ o? O* P^4 <f Month 6.0 5.0 4.0 N E ^ 3.0 Q O) 2.0 1.0 E. pacifica 12(9) X -544X-X J4(6)_ A 7(1) 0'^ «* • ^ ^ /* >* ^ cf cfi- ^ J Month Figure 2.6. The seasonal biomass patterns of euphausiids after correcting day samples (stars) by 20 and 2.5 times, respectively. Nighttime and crepuscular samples (squares) were not corrected. Lines join arithmatic averages. Note that 1 sample from T. spinifera data is excluded from the October mean because it is a statistical outlier. Also note that 23 of 50 total samples were collected during the daytime. 66 How reasonable are the magnitudes of the adjusted euphausiid biomass estimates? Some estimates of euphausiid biomass are available from the literature, but caution must be used in comparing them because of different species, locations, sampling gear, and seasons. However, as rough estimates of maximum biomass, the adjusted Eddy data 20-70 g WW m"2 compare with data in Simard et al. (1986) 15-133 g WW m"2; Sameto (1983) 2-315 g WW m"2; and Fulton et al. (1982) 1-80 g WW m"2. To sum up, within the Eddy region during the late spring/early summer it is apparent that: 1) net sampling for large euphausiids was conducted with an inefficient, small net primarily during daylight, 2) pelagic fish feeding data suggest high T. spinifera biomass, 3) T. spinifera populations consist primarily of large (> 15 mm) adults and thus high biomass. I conclude that it is not unreasonable to assume that T. spinifera biomass is underestimated by as much as 20 times during this period. It should be recognized that there are potentially large differences in catchability between euphausiid species. For example, E. pacifica is probably only underestimated by < 3 times versus as much as 20 times for T. spinifera using Bongo nets during the day. This result highlights the importance of not applying one night/day catch ratio to all euphausiid species (See also Brinton 1967). More precise biomass estimates of any euphausiid species could be obtained if sampling is conducted using more efficient sampling nets (e.g., BIONESS), and if sampling is consistently conducted at night (i.e. at least 1 h after sunset and 1 h before sunrise). To improve the accuracy of euphausiid biomass estimates and to determine if the hypothesized greater biomasses of T. spinifera does occur, more frequent sampling must be conducted during late spring/early summer. Simulated euphausiid versus empirical biomass patterns: The initial values of biological parameters discussed for euphausiids in section 1.5.3 generated a biomass pattern that did not calibrate very well to the corrected empirical seasonal biomass pattern discussed above, nor did it meet the late winter/early spring, or summer grazing demands 67 of the fish. Given this outcome, I considered three possibilities for improving the calibration of the simulation. One possibility is to reduce the estimate of euphausiid consumption by pelagic fishes. However, given that fish consumption and biomass are relatively well documented, the loss of euphausiids to fish grazing is probably of the right order of magnitude for the Juan de Fuca Eddy during the upwelling season (Tanasichuk et al. 1991; Ware and McFarlane 1994). A second possibility is that I underestimated euphausiid growth. This is particularly relevant in light of the fact that little is known about the biology of the dominant euphausiid, T. spinifera. Model euphausiids grow as a function of ingestion rate, prey availability, and gross growth efficiency (See Chapter 1). The maximum ingestion rate used (20% BWD) is already at the high end of the observed range, and thus should not be increased. Prey available to model euphausiids is restricted to diatoms and copepods, because little is known about the utilization or abundance of other prey like "detritus". By elimination, adjustment of the growth efficiency (GGE) term had the greatest potential for increasing model euphausiid growth. To determine how much to increase the euphausiid GGE, I initially obtained maximum growth rates in length for young adult euphausiids of four species (Table 2.3). I then converted the estimates of daily growth in length to growth in weight using published length-weight relationships for each species (Table 2.3). In this example, it is apparent that T. spinifera has a maximum growth rate about 2 times > E. pacifica, E. lucens, or N. australis. In subsequent simulations, I determined the GGE required to generate a model euphausiid average upwelling season growth rate that is similar to the measured maximum rate of 3.3% BWD (Figure 2.7). A euphausiid GGE of at least 25% is required to generate a seasonally averaged euphausiid growth rate of 3.3% BWD. Although a GGE of 25% generated an average growth rate that offset late winter fish predation, it could still not track the early summer empirical euphausiid biomass pattern, the empirical 68 Table 2.3. Measured maximum growth rates for adults of four species of euphausiid (See references). Growth in length is converted to growth in weight using the appropriate length-weight regressions. Also, estimated growth is calculated by assuming that starting length for each species is 12 mm, and that the euphausiids grew maximally for 30 days. References for Euphausia pacifica: Heath (1977), Bollens et al. (1992); Euphausia lucens: Stuart (1986); Nctyphianes Australis: Ritz and Hosie (1982); Thysanoessa spinifera: Fulton et al. (1982), R. Tanasichuk, DFO, pers. comm. (L: length in mm; W is weight as mg wet weight for E. pacifica and T. spinifera, and as mg DW for N. australis and E. lucens). Species E. pacifica E. lucens N. australis T. spinifera Measured growth (mm d"1) < 0 . 1 <0 .05 < 0 . 0 7 0.1-0.17 Length-weight regression ln(W) = -4.7 + 2.92*ln(L) W = 0 . 0 0 1 2 * L 3 1 6 W = 0 . 0 0 0 9 6 5 * L 3 0 4 5 ln(W) = -4.37+ 2.87*ln(L) Growth est. (%BWD) 1.8 1.3 1.6 3.3 69 10 9 8 7 CM E 6 £ 5 Q O) 4 3 2 1 _-_ -. M l ! I Standard ^M A \ „ 35% / v \ S • k • * I I I I I / I I zy~{-1 1 1 1 'On 20% ^ ' . , . . - • . • T • A, s s s ; i i i i i i i i i i i i i i i i i i i i i i V. •> ^^-JL .! I I s. I * • mm. I f <? ^ ^ & >? $ i& of O* ^ j Month Figure 2.7. Comparison of the standard run euphausiid biomass pattern (See Figure 2.8) with patterns generated using different euphausiid growth efficiencies (20%, 25%, 35%). Boxes indicate one standard error in sampling time and concentration. 70 hake ingestion rate pattern, or the fraction of euphausiids in the hake diet (Figure 2.7; see below). A GGE of > 40% is required for the simulated euphausiid biomass to calibrate closer to the observed biomass pattern. However, a GGE of this magnitude is unrealistic because it produces an average upwelling season adult euphausiid growth rate of 5% BWD, which is substantially higher than any rate recorded in the literature. The third possibility for improving the calibration of simulated euphausiid biomass to empirical data is to consider that euphausiid biomass is not only enhanced by high growth, but also by oceanographic mechanisms that bring euphausiids to, and accumulate them in, the Eddy region during the upwelling season (Ware and Thomson 1986; Simard and Mackas 1989; Ware and McFarlane 1994). The idea that physical oceanographic mechanisms accumulate zooplankton biomass in coastal regions is not new. Several studies in the coastal Benguela Upwelling system have indicated strong interactions between coastal current dynamics and euphausiid aggregations (See Barange and Pillar 1992). Roff et al. (1988) also hypothesized that euphausiid populations in the North Sea are enhanced from seasonal inputs from northern waters. It is also apparent that when oceanographic mechanisms breakdown, euphausiid aggregations "dissolve", although this process can be complicated by food-chain interactions (Brodeur and Pearcy 1992; Harris et al. 1992; Ware and McFarlane 1994). On the southern BC shelf, adult and juvenile euphausiids could be "imported" from upwelling regions along the continental shelf-slope interface (Simard and Mackas 1989), from deep water input via Spur Canyon (Freeland and Denman 1982), and/or from Juan de Fuca Strait outflow (Mackas 1992). Consequently, older stages of euphausiids would be retained in the Eddy region because of their diel vertical migratory behaviours (Simard and Mackas 1989), and because of the retentive circulation mechanisms associated with the cyclonic gyre (Freeland and Denman 1982). To explore the possible importance of euphausiid import, I linked this process to the seasonal pattern and intensity of upwelling as inferred from positive Ekman transport. 71 Thus, maximum import occurs during the week of greatest Ekman transport, and is proportional to remaining weeks during the upwelling season. Simulations demonstrate that euphausiid import results in a temporal biomass pattern that calibrates very well to the empirical data (Figure 2.8). In addition, average euphausiid growth rate remains around 3% BWD, and the seasonal hake ingestion patterns are satisfied (see below). Import and high average growth allows euphausiid biomass to increase four-fold from late April to early June. By mid-July however, there is a relatively rapid decline in empirical and simulated euphausiid biomass to roughly one-half that of the mid-June peak (Figure 2.8). This rapid decline is due primarily to predation by hake (see below). The empirical and simulated euphausiid data further indicate an increase in biomass from late July to early September (Figure 2.8). This increase is also linked to the effects of fish predation. As euphausiid biomass declines in the Eddy, hake begin to emigrate to the shelf break in search of more abundant supplies of prey (See Chapter 1). Thus, fish predation is reduced and growth and import continue to increase euphausiid biomass. The simulated and observed data indicate that there is roughly a two-fold decline in euphausiid biomass from early September until mid-November. This decline is due to the cessation of euphausiid import by the end of September and to slight negative growth associated with the decline in prey production. Both processes are a result of the relaxation in upwelling and concomitant increase in downwelling (see Figure 1.2). 2.2.4 Fish biomass patterns Simulated seasonal biomass patterns for hake, juvenile herring, and adult herring are presented in Figure 2.9. Hake biomass reaches a maximum of about 200 kt by the end of June and declined during July and August as a result of emigration from the Eddy region (linked to euphausiid biomass), and a result of the summery fishery. The decline in hake biomass during late summer is consistent with observations of the hake fishing fleet 72 «0 .£> »<£ *°* <r Month Figure 2.8. Simulated euphausiid biomass pattern calibrated to 1985-89 empirical data. Boxes indicate one standard error in sampling time and concentration. 73 200 180 160 «T140 o § 1 2 0 w 1 0 0 0) O O IUUQ1 40 20 0 -Juvenile ~ herring i i i i i i i i N Pacific \ hake ' . / \ Adult N herring ' v M , - - - - .*-*- \ - - - - K / ' ^ ' . I I I I I I. ' M l I I I I I. I l \ \ \ \ \ \ \ \ \ \ Month Figure 2.9. Simulated annual biomass patterns of juvenile herring, adult herring, and Pacific hake for the Eddy region during 1985-89. 74 moving further offshore in pursuit of the hake stock (see Chapter 1 and Ware and McFarlane 1994). By late August, about 100 kt of hake remained in the Eddy area, consistent with hydroacoustic estimates. Hake biomass continues to decline until late October, when no hake remain in the Eddy region. To further ensure the efficacy of the hake biomass pattern, I calibrated model output to empirical data describing the hake ingestion rate, and the fraction of euphausiids in the hake diet (Figure 2.10), while at the same time ensuring that the empirical euphausiid biomass pattern is maintained (Figure 2.8). Calibrating the model to these three variables generates the most realistic biomass patterns; simulations using only one variable result in very different patterns of hake and euphausiid biomass. Adult herring biomass increases in the Eddy region during April as they immigrate from inshore spawning grounds (Fig 2.9). In May and June adult herring biomass grows as a result of feeding on euphausiids. During July and August adult herring biomass declines to about 54 kt, primarily as a result of hake predation. Over the summer and autumn, an average adult herring growth rate of about 0.27% BWD balances mortality from piscivores. The average model adult herring growth rate compares well with that estimated by R. Tanasichuk (0.29% BWD; DFO, Nanaimo). Juvenile herring ingest about 2.5% BWD of copepods to balance mortality from hake, dogfish, and salmon, and to generate a copepod biomass pattern consistent with empirical data (Fig 2.3). 2.3 Model corroboration In section 2.2, I described the calibration of the trophodynamics model to the available empirical data. The relatively good agreement between the standard run model output and the empirical data is a measure of the validity of the simplifying assumptions used. To further support this conclusion, I corroborate the production dynamics of the standard 1985-89 run to observations made in other regions of the Coastal Up welling Domain. Because of the surprising lack of information describing production dynamics in A 75 4.0 a ID a 3.0 C 2.0 O IB CO <u O) c z 1.0 Q Q M | | | | | | | | | | | | | | | | | | ,1 | | I | | M I M I I I I I I I | | I I | | I I M I • \ \ \ \ \ \ \ \ \ \ \ \ ^ * * * o* ^  <r Month B So., to 2 0.8 'to 3 C3 3 <D §0 .6 •c O a o.5 o / X \ „ X ^< x \ M l M M I I 1 I I 1 1 0.4 ^ « * ^ ^ * * V>* ^ ^ <>* °* ^ O* Month Figure 2.10. Calibration of simulated hake ration (A) and fraction of euphausiids in hake diet (B) to empirical data (X's) from Tanasichuk et al. (1991) 76 the Coastal Upwelling Domain, I also corroborate model output using observations made in the Benguela Current upwelling system located off the west coast of South Africa. I chose the Benguela system for two reasons. The Benguela system is effectively divided into northern and southern regions, depending upon the duration of upwelling favourable winds. There is little seasonal variation in wind-induced upwelling in the northern Benguela, while there is seasonal variation in the southern Benguela (Pitcher et al. 1992). Thus, the physical processes operating in the southern Benguela are comparable to the northern Coastal Upwelling Domain. I also chose the Benguela system because it has a remarkably similar species complex, from plankton to fish, when compared to the Coastal Upwelling Domain. In particular, mesozooplankton biomass is dominated by large copepods like Calanus spp, while macrozooplankton biomass is dominated by the euphausiids, Nyctiphanes capensis and Euphausia lucens (Hutchings et al. 1991). In addition, fish biomass in the Benguela system is dominated (> 80%) by anchovies, hake, mackerel, and sardines; a situation not unlike the Coastal Upwelling Domain (see Figure 2 in Ware 1992). Three properties of the upwelling system's production dynamics are considered: 1) annual production, 2) production-to-mean biomass (P/B) ratios, and 3) the efficiency of transfer of production from one "trophic level" to another. I conclude section 2.3 with an annual budget of the trophodynamical and non-trophodynamical flows observed during the standard 1985-89 simulation run. 2.3.1 Production dynamics Simulated estimates of annual plankton and fish production are calculated by summing daily biomass growth increments before losses, and then converting to carbon using conversion factors discussed in section 1.5. The limitations of model assumptions on, and temporal variability in, the production estimates are discussed in Chapter 3. Here I corroborate the simulated production estimates as summarized in Table 2.4. 77 Table 2.4. Simulated annual plankton and fish production properties for the Eddy region during the period 1985-89. Transfer efficiency is calculated by dividing zooplankton or fish production by diatom production and multiplying by 100. Adult herring properties are calculated for 1 April to 15 November, while hake properties are for 1 June to 15 October (Cop: copepods; Euph: euphausiids). Property Production (P; g C m"2 y"1) Mean biomass (B; g C m"2) P/B ratio Transfer efficiency (%) Diatoms 332 2.7 123 . . . Cop 24.3 0.9 27 7.3 Euph 11.9 1.5 7.9 3.6 J. herring 0.8 1.6 0.5 0.2 A. herring 1.5 2.4 0.62 0.4 Hake .98 5.0 0.2 0.3 78 Diatom production: Simulated annual diatom production for the 1985-89 run is 332 g C m y . In upwelling regions, net phytoplankton generally constitute the largest fraction of total primary production (Legendre and Le Fevre 1992). For example, net phytoplankton contribute on average about 80% (CV = 15%) towards total primary production during the upwelling season off central California (Garrison 1976). Walker and Peterson (1991) also note that about 70% of total primary production could be attributed to large diatoms (> 10 |im) in the southern Benguela upwelling system. Thus, total primary production in the Eddy region, as estimated from simulated diatom production and the above observations, may range 415-475 g C m y"1. Surprisingly, I could find no published estimates of annual primary production for the west coast of Vancouver Island. However, Perry et al. (1989), using data collected during the 1970s for the Washington shelf, estimated annual total primary production to be 650 g C m"2 y . This estimate is likely high because the authors lumped temporally (and spatially) variable cruise data into seasons before estimating annual production (e.g., spring and summer CVs are 70% and 90% respectively). The 1985-89 Eddy simulation provides an alternative estimate of diatom production based on what is required to generate empirical zooplankton concentrations. Actual measurements of primary production are however eventually required to corroborate the simulated estimate. On a Domain-wide scale, it appears that primary production in the northern Domain is > estimates from the southern Domain (e.g., 150 g C m y ; Smith and Eppley 1982), while generally < estimates from the central Domain (e.g., 540 g C m"2 y"1 at 37°N; Chavez et al. 1991). These results are consistent with what is known about variability in characteristics of upwelling throughout the Domain. For instance, upwelling is about twice as strong in the south-central Domain compared to the northern Domain given the same alongshore wind stress; this is because the Coriolis force is higher in the north (Huyer 1983). Simulated primary production for the northern Domain is about 2-4 times < 19 estimates made for the southern Benguela system (700-1300 g C m"2 y"1; Brown et al. 1991). Primary production is higher in the Benguela system because of higher upwelling rates, and lower variability in upwelling favourable winds (Ware 1992). Additional corroboration of the simulated primary production estimate comes from an evaluation of daily primary production rates. The simulated daily production rate during the April to October upwelling season ranges from 0.2 to 2.9 g C m"2 d"1, with an average of 1.4 g C m d . Perry et al. (1989) noted for the Washington shelf that daily production during 1974-82, ranged from 1 to 4 g C m"2 d . In the southern Benguela, mean daily production rates are estimated to be 1.6 to 3.5 g C m"2 d"1 (Brown et al. 1991). Given an estimate of annual primary production, production-to-mean-biomass ratios (P/B) can be calculated. For the 1985-89 simulation, the annual P/B ratio is about 123:1, while for the upwelling seasonal the P/B ratio drops to 85:1 (due to higher average biomass). Annual phytoplankton P/B ratios range from about 114:1 to 153:1 in the southern Benguela (Brown et al. 1991). The simulated diatom P/B ratios translate into doubling times of 3-4 days, while Benquela phytoplankton double every 2-3 days. A comparison of these doubling times to Eppley's (1972) estimate of optimal phytoplankton doubling rates (see section 1.5.1), indicates that both simulated and Benguela phytoplankton are limited by nutrients and/or light (see also Brown et al. 1991). Simulated diatoms are about 10 times more productive during July to September than winter, but only 2 times more productive than April to June (Figure 2.11). These results are consistent with expectations concerning periods of greater light limitation in winter. Copepod production: Empirical estimates of annual zooplankton production in coastal upwelling regions are even more uncommon than annual primary production estimates. However, an independent first-order estimate of copepod production for the Eddy region can be derived using the technique described in Hutchings et al. (1991): copepod production = B*G*Gp , where B is the average annual copepod biomass, G is the number of copepod generations during the upwelling season, and G p assumes that 80 Diatoms Copepods Euphausiids Month Figure 2.11. Simulated monthly production to mean biomass ratios (P/B) for diatoms, copepods, and euphausiids. 81 production is 2-3 times the standing stock (Steele 1974, in Hutchings et al. 1991). Estimated annual average standing stock of copepods in the Eddy region during the 1980s is about 2.75 g DW m or 1.3 g C m"2 (from Figure 9 in Mackas 192). The average upwelling season lasts about 6-7 months (see Chapter 3), and neritic, temperate copepods have about 6-8 generations per upwelling season (assuming no food-limitation and mean temperatures of 11°C; Corkett and McLaren 1978), thus annual copepod production is calculated at 16-31 g C m"2 y"1. Simulated annual copepod production, 24.3 g C m y"1, falls in the middle of the estimated range. Simulated annual copepod production is about 1.2 to 1.6 times < copepod production estimated for the southern Benguela (e.g., 30-40 g C m y"1; Hutchings et al. 1991). Verheye et al. (1992) list daily copepod production estimates for the Benguela system. As expected the range of daily estimates is broad, but most are between 50-250 mg C m d"1; the simulated mean daily rate for the Eddy is 66.7 mg C m"2 d"1, and 130 mg C m d"1 during the upwelling season. Simulated annual and daily copepod P/B ratios are also lower than those estimated for the Benguela system (annual 28 y versus 30-40 y ' 1 ; 0.08 day"1 versus 0.1-0.2day"1 , respectively; Table 2.4). Higher annual and daily copepod production and P/B ratios in the Benguela are consistent with the observed higher phytoplankton production (see above). Euphausiid production: The standard run estimate of annual euphausiid production is 11.9 g C m"2 y"1, while the calculated P/B ratio is 8 (Table 2.4). The euphausiid P/B ratio initially seemed high, but perusal of the literature indicates that it is reasonable (Table 2.5). In fact, in the Benguela system, Hutchings et al. (1991) estimate euphausiid production at 6-20 g C m y" , with P/B ratios of 10-16. The relatively high euphausiid P/B ratios in the Benguela system are due to abundant food supply, prolonged breeding season, and low seasonality in water temperatures (Pillar et al. 1992). Fish production: Annual fish production in the model is about 100 times < diatom production and about 11 times < zooplankton production (Table 2.4). The simulated total 82 Table 2.5. Published euphausiid production to mean biomass (P/B) ratios. Euphausiid E. pacifica E. pacifica E. superba E. pacifica E. pacifica T. raschii M. norvegica T. longicaudata T. inermis N. australis E. lucens T. spinifera MEAN Location California Oregon Antartica Oregon BC Quebec Scotland Australia South Africa BC P/B 3 3 1.2 - 3 8.7 8.4 -9 .5 4 2.3 1.2 -11.6 1.3 - 4.2 13 -15 10 -16 8 5 - 7 Reference Lasker (1966) Mullin (1969)** Allen (1971)* Smiles and Pearcy (1971) Heath (1977) Berkes (1977a)** Mauchline (1977b)** Lindley (1978a)** Lindley (1980)* Ritz and Hosie (1982) Stuart and Pillar (1988) This study (w/o this study) * Cited in Stuart and Pillar 1988 ** Cited in Mauchline 1980 83 9 1 • fish production of 3.3 g C m" y" is generally comparable to other productive shelf regions like Georges Bank and the North Sea (2.0-4.0g C m"2 y"1; Jones 1984; Sherman et al. 1988). The simulated total fish production is also consistent with estimates made by Iverson (1990) for non-up welling coastal and oceanic regions. Iverson (1990) related fish production (FP) to phytoplankton production (PP; 50-300 g C m"2 y"1) using the equation: FP = -3.73 + 0.095*PP. Substituting the simulated primary production into Iverson's regression and assuming carbon is 6.75% of wet weight, then 1.9 g C m"2 y"1 of fish (and squid) is generated. Simulated herring and hake production is about 1.7 times > Iverson's estimate, as would be expected for a more productive coastal region. Compared too the r up welling regions, simulated fish production is about 5-9 times lower than estimates for coastal Peru (17-27 g C m"2 y"1; Jarre et al. 1990), and is similar to estimates for the southern Benguela (see Table 2.6). Fish production in the Coastal Upwelling Domain would be expected to be lower than other coastal upwelling regions because yield efficiency (catch divided by primary production) is about 2-3 times lower (Ware 1992). Yield efficiency is lower in the Domain because the intensity and duration of upwelling is about 3-4 times lower (Cushing 1978; Ware 1992). Simulated annual hake production contributed 30% to total annual fish production, while contributing about 60% to total biomass. The simulated annual hake production of 0.98 g C m' y" is similar to lower-end estimates made for hake off coastal Peru (0.9 - 2.1 g C m" y"1; Jarre et al. 1990). An estimate of the annual hake P/B ratio can be generated from the relationship developed by Banse and Mosher (1980) for temperate fish: P/B = 2.75 * W , where W i s weight in kcal. The average body weight of 6-8 year old hake on the southern BC shelf from catch data during the mid 1980s is about 900 g (La Perouse reports 1986-90; Leaman 1992). Assuming that 1 g WW is 6.75% carbon (C) and that 1 g C is 11.4 kcal (Pace et al. 1984), then a first-order estimate of hake P/B is 0.5. Jarre et al. (1991) used a hake P/B ratio of 0.3-0.35 in modelling studies of coastal Peru food-webs, while Shelton (1987) indicates that hake in the Benguela system have a P/B of 0.3. The 84 Table 2.6. Published estimates of phytoplankton (1°), zooplankton (2°), and fish (3°) production (g C m y ) in productive coastal non-upwelling and upwelling regions. Transfer efficiencies (%) for diatom-to-zooplankton (TE1_2) a n d f ° r diatom-to-fish ( T E ^ j ) are calculated as zooplankton or fish production divided by primary production times 100. NON-UPWELLING North Sea Georges Bank Bering Sea Bering Sea Scotian Shelf General shelf UPWELLING N Benguela S Benguela Peru 1953-59 Peru 1960-69 Peru 1973-79 Peru Canary BC S California Washington 1° 100 90 100 83 373 332 200 400 102 300 432 735 890 875 987 1225 730 332 343 650 2° 16 17.5 18 35 35 35 45 30 8 — 48 21 116 113 149 — 65.0 36.8 — 35.0 3° 0.6 0.8 0.8 0.8 0.5 0.5 . . . . . . 0.3 1.7 . . . 2.7 19.2 26.4 17.4 14.2 . . . 2.4 1.2 — T E l - 2 16.0 19.4 18.0 42.0 9.4 10.5 22.5 7.5 7.8 — 11.1 2.9 13.0 12.9 15.1 — 8.9 11.0 — 5.4 TEL3 0.6 0.9 0.8 1.0 0.14 0.15 . . . . . . 0.33 0.6 . . . 0.4 2.2 3.0 1.5 1.2 . . . 0.7 0.35 — Reference See Jones (1984) Cooney and Coyle (1982); Vidal and Smith (1986) Mills and Fournier (1979) Iverson (1990) Brown et al. (1991); Hutchings et al. (1991); Ware (1992) Jarre et al. (1990) Ware (1992) Mann and Lazier (1992) This study Ware (1992) Perry et al. (1989); Landry and Lorenzen (1989) 85 simulated hake P/B ratio of 0.2 is roughly 1.5-2.5 times < the above estimates, but is similar to annual instantaneous mortality rates of 0.2-0.27 estimated for hake on the southern BC shelf (Leaman 1992). In the model, 65% of the total annual herring production is generated by adults. Simulated juvenile herring production is underestimated because detailed growth dynamics are not incorporated. The calculated adult herring P/B ratio of 0.62 compares to a ratio of 0.8 obtained using Banse and Mosher's (1980) equation (see above; assumes that an average age 3-5 herring weighs 140 g WW). The adult herring P/B ratio is also about 2-4 times < estimates for other clupeoids off southern Benguela (e.g., anchovy P/B = 1.33, Armstrong et al. 1991) and Peru (anchovy and sardine P/B = 1.8-2.5; Jarre et al. 1991). 2.3.2 Transfer efficiency The flow of production among dominant organisms of a pelagic system can be summarized using ecological or transfer efficiency (Parsons et al. 1984). Transfer efficiency is simply calculated by dividing simulated annual zooplankton or fish production by simulated annual diatom production (Table 2.4). In general, transfer efficiency declines with increasing rate of primary production. Cushing (1971) first described this relationship for upwelling regions (Figure 2.12). I have developed a similar plot of more recent estimates of primary and secondary production for continental shelves in upwelling and non-upwelling regions (Figure 2.12; Table 2.6). Both Cushing's and the present-day data exhibit positive correlations between primary production and secondary production (r= +0.28 and r= +0.65, respectively). However, the slope of the present-day production-transfer relationship is about one-quarter that of Cushing's estimate, while the mean transfer efficiency of present-day upwelling systems is about one-half (10% versus 18%, respectively). These results suggest that the transfer efficiency may be less variable among productive oceanic regions, than first suggested. Note also that there are large differences in transfer efficiencies within 86 ^ 6 0 -^ 50 o C 40 .0 O •S 30 a. C 20 1 X1 0 o o N o • • • _ .— • • • • 1 • A •v A • • 1 1 v * 1 I • -A — • v . • -* 1 1 30 25 S5 2 0 £>* C d> 15 10 Primary production (g C/m3/y) 3= fl> •*-W c Primary production (g C/m /y) Figure 2.12. Zooplankton production (solid line and squares) and transfer efficiency (broken line and triangles) plotted against primary production. Panel A: data from Cushing 1971; Panel b: data from Table 2.6. 87 upwelling regions in both data sets. These differences result from variability in interactions between the intensity and duration of upwelling, width of the continental shelf, and food-web coupling (see Ware 1992). For example, transfer efficiencies in the Benguela system range from 3 to 11%, which can be related to the seasonality in upwelling favourable winds (Table 2.6; Pitcher et al. 1992). Present-day estimates of annual primary production and fish production are also positively correlated (Figure 2.13). Although there are fewer data available, the phytoplankton-fish transfer efficiency is positively related to primary production. The simulated mean annual diatom-to-fish production transfer in Figure 2.13 is 0.96%, which corresponds to an average empirical transfer of 0.98% for all data in Table 2.6, and to 1.7% for upwelling systems only. The combined positive primary-tertiary transfer correlation and negative primary-secondary transfer correlation (see above), implies that the major uncoupling in system production transfer of the coastal food-web occurs at the primary-secondary level (See Chapter 5). 2.3.3 Annual mass balance budget An annual budget of the trophodynamical and non-trophodynamical flows in the standard model run is presented in Figure 2.14, and discussed below. Diatoms: Of the total estimated annual diatom production (4933 t km"2), about 53% is transported from the surface layer of the Eddy to offshore regions. The result that the largest fraction of diatom production is lost to export is consistent with observations made in several coastal upwelling systems. Walsh (1981) for example, indicates that between 50-70% of the primary production off Peru is exported to slope regions. Pace et al. (1984) also determined from a simulation model that about 50% of annual primary production is lost (exported) from continental shelves to become detritus. Finally, Ware (1992) estimated that 70, 83, and 90% of the primary production is transported offshore in the Peru, Benguela, and southern California upwelling systems, respectively. Although simulated 88 30 £ 2 5 C\l 20 O CD w 1 5 C _o § 10 o a 5 W £ o --- • • -'Z/ / ^ A • • • ^ ^ 40" ^^ ^" ^ ^ ^ -"** ^^ *>" ^ * : A * • • I I i I ^ ^ / -• _ *°-3.5 3.0 2.5 O 2.0 .2 - 1.5 - 1.0 0.5 I 0.0 Primary production (g C/m /y) Figure 2.13. Fish production (solid line and squares) and transfer efficiency (broken line and triangles) plotted against annual primary production. See Table 2.6 for values, 1.9 6.3 Dogfish 7s 1 2.0 42.3 1.3 12.2 A. herring 43.7 Hake 125 Euphausiids P = 179, B=22 Diatoms P = 4929 B = 40 706 1299 tf 84 Man 8.1 Salmon 2.0 J. herring 64 Copepods P = 368 B = 13 2924 Export 220 Figure 2.14. Flow diagram of the standard 1985-89 model run. Flows are expressed in tonnes/km2/y. Note that the size of the plankton boxes are proportional to their annual production (P:production; B: mean annual biomass). 00 90 diatom export is similar to the first two studies, it is about half of what Ware (1992) estimated for the southern Domain (53% versus 90%, respectively). This result is consistent however with reduced upwelling rates and greater seasonality in Ekman transport in northern areas of the Domain (Huyer 1983). The simulated sinking loss of 7% of diatom production is about 2 times > estimates made for the Benguela system (3.7%; Brown et al. 1991). Higher sinking losses of diatoms in the northern Domain would be consistent with observed lower upwelling rates, and subsequently greater nutrient limitation (see above). Model zooplankton graze about 40% of diatom production, which is consistent with estimates from the literature (Figure 2.14). Pace et al. (1984) also indicate that, for various simulated production scenarios on continental shelves, about 50% of primary production would be grazed by zooplankton. Estimates of between 5-50% (mean: 25%) of phytoplankton production is estimated to be grazed by copepods in the southern Benguela depending upon season, state of upwelling, etc. (Verheye et al. 1992). Pillar et al. (1992) estimate that euphausiids consume about 1-3% of primary production, about 5 times < than the simulated estimate. In the northern Domain, Landry and Lorenzen (1989) noted that grazing by copepods and euphausiids accounted for about 60% of the total zooplankton grazing impact, which removed about 20% of the primary production on the Washington mid-shelf. Copepods: Of the total copepod production, about 23% is grazed by euphausiids (Figure 2.14). In addition, about 30% of this total is grazed by euphausiids in April /May, but only half as much in the autumn. Pillar et al. (1992) similarly estimate that Benguela euphausiids annually consume between 21-35% of annual mesozooplankton production. The authors also note that between 7-60% of copepod biomass is removed by euphausiids depending upon the latter's "swarming" behaviour. Juvenile herring consume about 16% of the total annual copepod production. Of this amount, 10% is consumed by herring in spring, while 70-90% is consumed in October/November. 91 Euphausiids: Euphausiids lose 70%, 25%, and 5% of their annual production to hake, adult herring, and dogfish grazing, respectively. Fish graze about 270 kt of euphausiids, with simulated consumption to mean biomass ratios (C/B) of about 1.7 for hake, 1.1 for adult herring, and 1.0 for dogfish. The simulated hake C/B is slightly < estimates made for hake off Peru (1.8-2.2; Jarre et al. 1991). The herring C/B ratio is about 6 times < that estimated for clupeoids off South Africa (6.9; Armstrong et al. 1991), while about 2 times > than estimates made in the Bering Sea (0.7; Laevastu et al. 1982). The simulated dogfish C/B ratio is about 2 times < estimates made for dogfish in coastal BC (2-2.5; Geen and Jones 1977). Although the simulated and observed comparisons of fish C/B ratios are consistent with expectations concerning system production, actual comparisons are dubious because most of the above studies used simplifying assumptions concerning seasonal variability in fish ingestion rates, diet composition, and biomass. The simulated estimates of euphausiid consumption are first-order because several "members" of the pelagic "food-web" have been ignored. We know that sablefish, Pacific cod, Pacific mackerel, and various adult and juvenile salmon, among others, feed on euphausiids in shelf regions of the Northern Domain (e.g., Brodeur and Pearcy 1992). Although these fish species comprise a relatively small biomass of fish compared to those on the BC shelf and those included in the model (e.g., Ware and McFarlane 1994), they may still consume a reasonable proportion of euphausiid production. If we also consider losses of euphausiids to sea birds (Mackas and Galbraith 1992), baleen whales, and to invertebrate predators (e.g. squids, chaetognaths), the estimate of euphausiid production would have to be raised to satisfy demand. The fundamental question however still remains: what mechanisms or processes result in such extremely high euphausiid production in the Juan de Fuca Eddy to satisfy the high trophic demand? 92 2.4 Model sensitivity The calibration and corroboration of the standard run in sections 2.2 and 2.3 provides confidence and evidence that the trophodynamics model is capturing the essential features and dynamics of system production, as we presently know them. However, it must be remembered that several critical details of the Juan de Fuca Eddy upwelling system are not known with certainty. Thus, it is necessary to evaluate the influence of model parameters or processes on simulated output, and to highlight areas where additional data are required. The sensitivity analyses are described in three major sections according to the level of uncertainty in parameter values or processes. Initially I describe the influence of observed variability in starting concentrations of plankton and fish on model output. I then evaluate the variability in less well known plankton, environmental, and fish parameters on model output. Finally, I evaluate model sensitivity to assumptions concerning the timing of the arrival and biomass of Pacific hake. The output that I have used as an indicator of model sensitivity is the annual production of plankton and fish, and the empirical seasonal biomass patterns of the copepods and euphausiids (see Figures 2.3 and 2.8). To quantify the response of zooplankton biomass to parameter or process variability I use the negative log-likelihood function: J M SEt which minimizes the differences between the mean simulated (Sj) and observed zooplankton biomasses (Oj) at each of 8 sampling periods (i; See boxes in Figure 2.3). The weighted sum of squares approach includes the standard error of the observed biomass estimates for each period (SE;), and thus gives more "weight" to those periods better sampled (Smith et al. 1990). To calculate the mean simulated zooplankton biomass (Sj) for period (i), I include plus or minus seven days around the mean sampling date. 93 2.4.1 Starting concentrations of state variables I tested the sensitivity of annual production to starting concentrations of state variables because it has been shown in several ecological modelling studies that starting conditions can influence trophodynamical interactions (e.g., Evans and Parslow 1985). The ranges of possible starting conditions were determined by calculating standard errors in winter diatom biomass (data from Landry et al. 1989), and in winter zooplankton biomass (data from D. Mackas, IOS), and coefficients of variation in herring, dogfish, and salmon biomass (data from Ware and McFarlane 1994; see also Chapter 3). I leave an evaluation of hake biomass variability to section 2.4.3. Perturbations in starting concentrations of state variables have relatively small effects (< 10% change) on diatom and euphausiid annual production (Table 2.7). In addition, diatom and euphausiid annual production shows 3-4 times less variability than copepods to perturbations in starting concentrations. Copepod production is most influenced by changes in adult herring biomass, which results in changes in euphausiid predation effects (Table 2.7; See also Chapter 4). Adult herring in turn are most sensitive to changes in their own biomass, while hake production is relatively sensitive (8-16% change) to variability in euphausiid and herring starting biomasses (Table 2.7). Interestingly, annual hake production decreases with increasing herring biomass because more herring consume more of the euphausiid spring biomass, and thus hake immigrate from the Eddy region earlier than if more euphausiids were present . Sensitivity of the zooplankton seasonal biomass patterns to perturbations in starting concentrations was investigated using the negative log-likelihood function ("f" values). In the standard run, both simulated copepod and euphausiid biomass are close to observed biomasses in 7 of 8 sampling periods, and thus had similar f values (Table 2.8). The relatively high zooplankton f values are a result of the simulated euphausiid biomass being > the observed May biomass, while simulated copepod biomass is > the observed November copepod biomass (Table 2.8; see Figure 2.3 and 2.8). Perturbations in starting 94 Table 2.7. Percentage change in annual plankton and fish production from standard run values (see Table 2.4), using listed starting values of state variables. (Changes < 1.0% are denoted as 0; kt: thousands of tonnes; DW: dry weight). Variable Diatoms (mg Chla m"3) M Copepods (mg DW m"2) it Euphausiids (mg DW m"2) 1! J. herring (kt) it A. herring (kt) M Dogfish (kt) II Salmon (kt) H Start Value 0.85 0.15 0.90 0.50 2.6 1.4 35.8 19.3 71.5 38.5 40 10 22 5.5 % CHANGE FROM STANDARD RUN ANNUAL PRODUCTION Diat -1.0 0 -1.7 0 -3.6 0 2.1 -4.2 -5.1 0 -3.0 0 0 1.5 Cop 0 -13.2 4.9 -16.4 -4.1 18.5 -7.8 12.8 21.4 -3.7 14.4 1.2 2.5 4.9 Euph 0 1.7 0 2.9 8.9 -8.0 5.3 -2.5 0 4.3 1.3 2.5 0 2.5 Herr 0 0 0 2.0 0 -2.0 2.1 0 37.0 -38 -4.8 2.7 -10 6.2 Hake 0 4.2 4.2 4.2 12.5 -8.3 4.2 0 -8.3 16.7 0 4.2 4.2 0 95 Table 2.8. Standard run log-likelihood "f" values for copepod and euphausiid simulations versus mean "f" values obtained by perturbing starting concentrations of state variables, and by perturbing abiotic, plankton, and fish parameters (see Table 2.9). Standard run Starting values Abiotic parameters Diatom parameters Copepod parameters Euphausiid parameters Fish parameters Copepods 9.6 44.4 19.7 46.9 55.9 36.4 24.8 Euphausiids 10.72 10.98 11.01 10.82 10.78 10.84 10.72 96 concentrations result in a mean copepod seasonal biomass pattern with a 5 times greater mean f value than the standard copepod run. The variability occurs primarily during June and November. In contrast, the perturbed and standard run euphausiid biomass patterns have similar mean f values (Table 2.8). Long-term simulation: Several published modelling studies evaluate output obtained after many (2-100) consecutive simulations, because the influences of starting values are "removed" (e.g., Frost 1987). To evaluate this effect on simulated production estimates, I ran the model for 50 consecutive years. I assumed a priori that if annual plankton production and year-end biomasses remained within +/- 5% of values simulated after one year, then the model was producing "stable" or steady output (See Distefano et al. 1967). Both annual diatom and euphausiid production and year-end biomasses vary by < +/- 5% over 50 consecutive model runs (Figure 2.15). Copepods also show little variation in annual production or in year-end biomass for the majority of years, but occasionally annual production "jumps" above or below the +/- 5% range. Closer examination of the output indicates that over the 50 y simulation, copepod and diatom annual production are out of phase (Figure 2.15). This result is consistent with observations that Michaelis-Menten type plankton feeding functions are depensatory, and result in "limit-cycle" oscillations (DeAngelis et al. 1989). Two important points however, are worth making. Mullin et al. (1975) could not statistically detect significant differences between rectilinear, Michaelis-Menten, or other curvilinear functions (e.g., Ivlev) for describing copepod ingestion. Thus, the choice of the plankton functional response is justified. The second point is that when the model is forced by empirical environmental variability (See Chapter 3), the effects of the feeding functions become secondary in importance. In other words, the choice of feeding function may be important in determining model output in a constant environment (DeAngelis et al. 1989), but when real environmental variability is used in the simulations the former is "obscured" by the latter. For an overview of the relative importance of biotic versus abiotic processes on model output see Chapter 4. A 400 ^ 3 5 0 E 300 O 2 2 5 0 C o •g 200 3 •D 2 150 a o 100 a 5 50 400 - s 3 5 0 E 300 O 2 2 5 0 •g 200 3 2 150 a E o 5 100 50 DIATOMS EUPHAUSIIDS 16 C o 3 O (0 3 £ a 3 UJ Q l I I I I I I I I I I I I I I I I I I I I | Q 1 6 11 16 21 26 31 36 41 46 51 Simulation Year B DIATOMS COPEPODS / Q l I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 1 6 11 16 21 26 31 3? 28 ?4 20 16 12 8 ^ ,^ >« CM E o O) * • " • * c o u 3 •o o h-o. •a o a. d> a 0 O I I I I I I I I I I I I I I I I I | Q 36 41 46 51 Simulation Year Figure 2.15. Plankton production produced over 50 consecutive simulations. Panel A: Diatom and euphausiid production with +/- 5% first year boundaries. Panel B: Diatom and copepod production, showing oscillations over time. 98 2.4.2 Parameter perturbations The set of parameters described in Tables 1.1, 1.2, and 1.3 is undoubtedly only one of some unknown number of possible sets that could generate the standard run output. This is because many parameters have a wide range of values that depend upon the choice of model components (e.g., copepod body weight), and because some parameters are presently unquantifiable (e.g., diatom export rate). However, given that the parameter set generates output that is consistent with empirical data (See section 2.3), it is useful to evaluate some relative level of parameter variability. Parameters are frequently varied one at a time by plus or minus 10% of nominal values, and simulated output is compared to the standard run (Swartzman and Kaluzny 1987); I use this approach. To facilitate the overview of model responses to parameter variability, I present the coefficients of variation in annual production and mean annual biomass generated from perturbations of individual parameters lumped together into general groups (see Table 2.9). Diatoms: Annual diatom production is 2-3 times more variable to perturbations in abiotic parameters than to perturbations in the other three parameter groups (Table 2.10). Within the abiotic parameter group, a 10% increase (decrease) in upwelling rate produces the greatest change in annual diatom production (+5% and -12%, respectively). The diatom biomass pattern remains relatively similar irrespective of upwelling rate variability (Figure 2.16). Copepods: Annual copepod production and mean biomass is about 2-4 times more variable as a result of changes to copepod parameters than to changes in other parameters (Table 2.10). Within the copepod parameter set, a 10% increase (decrease) in copepod maximum gross growth efficiency results in a 48% (40%) increase (decrease) in annual copepod production (Figure 2.17). Perturbations in copepod parameters produce the highest mean log-likelihood f value, with the 4th (June) and 6th (August) sampling periods exhibiting the greatest deviations from observed biomasses (Table 2.8). Variable copepod GGE is responsible for the large differences between simulated and 99 Table 2.9. General groupings of parameters evaluated in sensitivity analyses. Simulations are conducted by perturbing one parameter at a time by plus or minus 10% of its nominal values (See Tables 1.1, 1.2, 1.3). Abiotic parameters Upwelling rate N below mixed layer Summer mixed layer depth Winter mixed layer depth max sunshine Diatom parameters Respiration rate Export rate N half-saturation constant Max sinking rate Copepod parameters Average body weight max GGE max ingstion rate export rate availability to euphausiids Euphausiid parameters Average body weight max GGE max ingestion rate import rate Fish parameters max ingestion rates prey half-saturation constants prey thresholds max growth efficiencies 100 Table 2.10. Coefficients of variation (%) in annual production and mean annual biomass for diatoms, copepods, euphausiids, adult herring, and hake resulting from individually perturbing parameters in the general groups by plus or minus 10% of nominal values (see Table 2.9). Abiotic param. Diatom param. Copepod param. Euphausiid param. Fish param. PRODUCTION Diatom Copepod Euphausiid Herring Hake 6.4 17.5 4.1 0.6 5.7 2.3 8.9 1.6 0.7 2.7 2.0 27.7 5.4 1.1 9.0 1.6 13.0 9.5 1.1 12.2 2.1 2.2 1.2 2.7 4.2 BIOMASS Diatom Copepod Euphausiid Herring Hake 2.9 12.5 0.8 0.2 2.6 2.6 6.3 0.7 0.2 1.2 4.5 23.5 1.6 0.4 3.7 2.7 12.0 2.4 0.7 5.8 2.7 3 1.3 .9 .7 101 20 CO E sz O O) E ~ Standard "' High upwelling ' Low upwelling Figure 2.16. Response of simulated diatoms to hioher (4 4 m H-I , , „ H Xf m ' UPWe'"n9 rat6S C ° m P a r e d t0 »• « ^ ™ (40nd A I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I N N N . N \ N \ \ K \ \ N ** <? ** J ^ f & jfi cf O* ^  </ Figure 2.17. Response of simulated copepods to higher (44%) and lower (36%) growth efficiency compared to the standard run (40%). 103 observed summer copepod biomasses. Euphausiids: Annual euphausiid production and mean biomass is about 2-3 times more variable given perturbations to euphausiid parameters than to changes in any other parameter (Table 2.10). The parameter generating the largest change in annual euphausiid production or mean annual biomass is the maximum euphausiid growth efficiency. See section 2.2 and Figure 2.7. The standard run log-likelihood f value for the euphausiid biomass pattern is slightly higher than for the copepods (10.7 versus 9.6, respectively). The majority of the variability in the standard euphausiid run comes from overshooting the observed May biomass (See Figure 2.8). On the whole, the euphausiid seasonal biomass pattern is less variable than the copepod's pattern to perturbations in any parameter (f ranges 10.7 to 11 versus 9.6 to 55.9, respectively; Table 2.8). Fish: Both adult herring and hake annual production and mean biomass are 2-4 times more variable when euphausiid parameters were perturbed, than when any other parameters were perturbed (e.g., euphausiid GGE; Table 2.10). 2.4.3 Variability in hake migration There are at least two characteristics of hake migration that may possibly impact plankton and resident fish production: hake arrival time and migrating hake biomass. The arrival time of hake to the southern BC continental shelf is not known precisely. However, because of life history traits like spawning times and swimming speeds (e.g., Baily et al. 1982), and initiation of northern commercial fisheries, we can generate first-order estimates of hake arrival. Hake can arrive on the BC shelf as early as the first week in May and as late as the last week in June (See section 1.5.4). In the simulations, I varied the hake arrival times between these two possible boundaries on a weekly basis. Variable arrival time of hake and their effect on simulated euphausiid biomass is shown in Figure 2.18. The model predicts that if hake arrive in early May, euphausiid Hake 250 200 (B » 150 O (0 100 0> £ 50 STANDARD \ LATE EARLY N. •  I.I I l-l I I I I I .I I I I I H I I I. \ \ \ \ , \ \ N \ \ , \ .N \ / 0 f f J> £ V f f f J> f 16 14 12 «M 10 E £ 8 Q o> 6 Euphausiids --— -I I I I I / LATE / ^ / / / / STANDARD f\ i i i i i i i i i M i I I i i i i i i i \ \ \ I I \ I \ t .V i i i i i i i i i EARLY "^Ll* • • • *^  ^ — 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 / <* f / / / ^V *f ^ <f* / ^ Figure 2.18. Response of simulated euphausiids to early (dots) and late (dashes) hake arrival, compared to the standard run (line 105 biomass will only slowly increase after hake began to emigrate from the Eddy to the outer shelf in July/August. Alternatively, a late arrival of hake at the end of June results in a 'build-up' of euphausiid biomass during June, and allows hake biomass to remain relatively high until early September (Figure 2.18). Simulated hake arrival time also generates large variability in both euphausiid and copepod annual production (Figure 2.19). Late hake arrival results in the greatest increase in euphausiid production (+40%) because of a reduction in predation on euphausiids in early summer. On the other hand, late hake arrival negatively impacts copepod annual production indirectly via the effects of euphausiid predation. Copepod production increases when hake arrival dates are earlier than the standard run arrival (June 1; Figure 2.19). Variable hake arrival does not significantly influence annual diatom production (Fig 2.19). Adult herring have a similar response to hake arrival as the euphausiids (not shown). Late hake arrival generates up to 12% more annual herring production. The increase in production is due to a reduction in hake predation, and higher growth from greater prey (euphausiid) production. Hake biomass: I estimate the variability in the hake biomass arriving on the southern BC shelf using a temperature-hake biomass function derived by Ware and McFarlane (1994; see Chapter 3). The authors found that warmer June/July SSTs result in more hake on the southern BC shelf. The minimum and maximum average June/July water temperatures observed during the 1980s would result in an estimated 84 and 336 kt of hake entering the Eddy region, respectively. This represents a +/- 60% change in hake biomass from the standard run (210 kt). The two extremes in hake biomass considered and the responses of the simulated euphausiids are shown in Figure 2.20. As expected, more hake result in a lower seasonal euphausiid biomass pattern in summer. Note that a proportional decrease in hake biomass results in substantially more euphausiids. This is discussed below. 50 C .2 40 o 3 •o o 1 _ a. c o c a c O) c (0 30 20 10 •10 -20 •30 « - 4 0 -50 L~" ^ " • • « * ^ _ * * . , EUPH> I_j£_*»* UJSIIDS '' ^ ' •^ DIATOMS - J-^ ^ * / COPEFODS . — —" ' • ' • # ^ ^ & $ ** ^ 4? Arrival date of hake Figure 2.19. Simulated effect of variable hake immigration date on annual plankton production. Changes in production are calculated as % deviations from standards listed in Table 2.4. Hake 300 / # f f / / ^ ^ 4 / / j* Euphausiids / ff ^V / *>v f 4 # / o*c Figure 2.20. Response of simulated euphausiids to higher (dashed line: 315 kt) and lower (dots: 105 kt) hake biomass compared to the standarad run (line). 108 Annual euphausiid production is negatively related to changes in hake biomass, while copepod production is positively related and diatoms are relatively unaffected (Figure 2.21). The magnitude of change in euphausiid production caused by hake biomass variability is asymmetrical. For example, hake biomass > 210 kt resulted in a relatively narrow range of annual euphausiid production (a 0-10% decrease). When hake biomass is < 210 kt the range of euphausiid production is broad (0-35% increase; Figure 2.21). The asymmetrical response in euphausiid production is undoubtedly due to interactions between the non-linear hake emigration function and euphausiid biomass. It may be expected that certain combinations of hake arrival time and biomass would occur in nature. For instance, warmer oceanic conditions, due to increased advection of warm southerly offshore waters northward along the coast, would result in more hake arriving earlier on the southern BC shelf (Beamish and McFarlane 1985; Ware and McFarlane 1994). Conversely, cooler coastal oceanic conditions would result in fewer hake arriving later. The model indicates that the combinations of the two hake migration characteristics are additive. Thus under cool oceanic conditions, euphausiid annual production would increase by 80%, copepod production would decrease by 57%, diatom production would remain similar (+3.3%), and herring production would increase by about 50%. It is likely that other limiting factors may influence euphausiid production, such as increases in other predators or density-dependent effects. Under warm conditions, annual euphausiid production would decrease by 6%, copepods would increase by 34%, diatoms would remain similar (+2.2%), and herring would decrease by 12%. The hake emigration function acts to "stabilize" the response of the euphausiids. DIA t TOMS , EUPh X j IAUS4 n DS • ^ v » / . , '•-, CC PEPCIDS * , + • , ^ C o +* o 3 o 1_ a c o •fcrf J* c <0 a c • M 0) O) c (0 £ u 30 20 10 0 -10 -20 -30 * N# N* N* N# N# ^ ^ ^ £* ^ ^ # Hake biomass (kt) Figure 2.21. Simulated effect of variable hake biomass on annual plankton production. Changes in production are calculated as % deviations from standards listed in Table 2.4. CHAPTER 3: THE INFLUENCE OF OCEAN CLIMATE ON COASTAL PLANKTON AND FISH PRODUCTION 3.1 Introduction Fishery oceanographers have long been interested in the interannual variability in properties of coastal pelagic fish, like production or recruitment (e.g., Ryther 1969; Cushing 1971). Fish properties are frequently and directly related to oceanic properties (e.g., sea surface temperature) because these time series are readily available. However, relationships between coastal ocean climate and fish properties frequently breakdown over time because they are complex, and often nonlinear (Cury and Roy 1989; Mann 1993). The complexity and nonlinearity in the relationships between oceanic and fish properties are primarily due to the interannual dynamics of the intermediate linkage, planktonic prey (e.g., Cushing 1990). Fishery oceanographers seldom have access to descriptive long-term data sets of plankton properties because the relative costs associated with frequent sampling of highly spatially and temporally variable organisms make extensive study infeasible. Of the longer-term plankton studies that have been conducted in the Coastal Up welling Domain, only a few have evaluated interannual changes in phytoplankton populations (e.g., Tont 1987; Lange et al. 1990). These studies have been primarily conducted off southern California, and have determined that interannual variability of diatom populations is due to a combination of changes in local upwelling and in the circulation patterns of major water masses. Most studies evaluating interannual variability in Coastal Upwelling Domain zooplankton, have used the extensive California Cooperative Oceanic Fisheries Investigations (CalCOFI) data set. Colebrook (1977), for example, evaluated interannual changes in the biomass of zooplankton off southern California from 1955-59. He found that zooplankton biomass was lower in warmer compared to cooler years, and attributed the differences to changes in the strength of the California Current and in coastal upwelling. Several studies further demonstrated that the strong interannual variability in zooplankton off California was most highly correlated with interannual variability in mass transport via the California Current from the north (Bernal 1981; Chelton et al. 1982; Roessler and Chelton 1987). In the northern Domain, relatively few studies have evaluated interannual variability in zooplankton biomass, and those that do are relatively short-term. For instance, Peterson and Miller (1975) evaluated interannual changes in zooplankton species composition off Oregon in relation to changes in upwelling from 1969 to 1972. Zooplankton biomass was found to be lower in warmer years, and attributed to reduced upwelling. Landry and Lorenzen (1989) list in their Table 5.1 several additional short-term zooplankton studies off Washington and Oregon coast. Interannual changes in zooplankton off southern Vancouver Island have been studied from 1985 to 1992, by Mackas (1992, 1994). The author determined that there were roughly 3 to 10 fold variations in zooplankton biomass during the 7 y period, but no single time scale or mode of variation was shared by all taxa. In addition, Mackas (1994) surmised that the interannual changes in zooplankton biomass were associated with variability in water properties and currents forced by the Aleutian low pressure system. Although few studies describe temporal variability in Coastal Upwelling Domain plankton, even fewer studies have attempted to hindcast properties using theoretical relationships between plankton and the coastal oceanic environment (See Evans and Pepin 1989). Smith and Eppley (1982), for example, used relationships between phytoplankton, water temperature and day length to hindcast primary production in the southern California Bight, over the length of the abiotic time series. Other well known relationships between plankton and the coastal environment include the seasonal transport of near-shore surface layer biomass seaward (Chavez et al. 1991), and the enhancement of production because of intermittent inputs of subsurface nitrogen to the surface layer 112 (Legendre 1990). The influences of coastal upwelling on plankton growth or mortality may ultimately depend upon the timing of the transition from downwelling (winter) to upwelling (summer) conditions (Strub and James 1988), the duration of the upwelling season (Wyatt 1980), the intensity of upwelling (Cury and Roy 1989), and/or the seasonal pattern (frequency) of upwelling. Coastal upwelling in turn, is primarily influenced by the direction and intensity of alongshore winds, which is linked to the position and intensity of atmospheric pressure systems. For instance, the North Pacific High pressure system typically ranges from about 28°N and 130°W in February to about 38°N and 150°W in August. As a result of the average location of the Pacific High, wind stress between 40°N and 50°N is equatorward from about April to October. For the remainder of the year, alongshore winds in these northerly latitudes are generally poleward and unfavourable for upwelling (Huyer 1983; Strub and James 1988). Given that there are well defined effects of atmospheric conditions on coastal upwelling, and the latter on the plankton, it would be insightful to evaluate the temporal relationships among these processes, and thus ultimately their influence on pelagic fish production. The main objectives of this chapter are to use the Eddy trophodynamics model to: 1) hindcast plankton, and subsequently fish production, as estimated from relationships with observed seasonal patterns in upwelling, water temperature, and solar radiation, from 1972 to 1990, 2) compare model output to empirical data describing plankton and fish properties, and 3) generalize about the production dynamics of plankton and fish in light of interannual coastal ocean climate variability. 3.2 Methods Recall that the trophodynamics model is "forced" by empirical seasonal patterns in upwelling, sea surface temperature, and solar radiation. I have calculated average weekly values of the three environmental functions for each year from 1972 to 1990, using 113 methods discussed in Chapter 1. The trophodynamics model also requires yearly data describing hake biomass/catch and juvenile/adult herring biomass in the Eddy region. For each year during 1972 to 1990, hake biomass in thousands of tonnes (kt) is estimated using a temperature/biomass function derived by Ware and McFarlane (1994): Hake biomass (kt) = SST * 196.428 -2231.771, where SST is the sea surface temperature averaged over June and July. This function results in higher hake biomass in the Eddy region during warmer summers. For simplicity hake biomass is assumed to never be < 100 kt; 1974-1977 and 1984 had this minimum hake biomass. Hake biomass removed by the fishery each year is taken from Leaman (1992; Table 3.1). Interannual variability in the biomass of adult herring occurring in the Eddy region is determined from data in Schwiegart et al. (1992). I assume that 100% of adult herring (age 3 and older) from the Barkley Sound stock and 25% of adults from the southern Strait of Georgia stock occur in the Eddy region during summer. In addition, juvenile herring biomass (ages 1 and 2) in the Eddy region is estimated as a combination of 1.9 times Barkley Sound adult biomass and 1.0 times southern Strait of Georgia adult biomass (D. Ware, Pacific Biological Station, Nanaimo, pers. comm) (Table 3.1). The abiotic patterns and estimated fish biomasses were entered into separate models for each year. Subsequently, a simulation was conducted by running the 1972 model, and then using parameter values at the end of the 1972 simulation as starting conditions for the next model year (1973), and so on until the 1990 simulation was completed. Starting (1972) conditions, other than abiotic patterns and fish biomasses, were taken from the 1985-89 standard run. 3.3 Results This section initially describes the annual production estimates for plankton and fish generated from a continuous simulation from 1972 to 1990. I then evaluate the 114 Table 3.1. Values of fish biomass and catch used in the Eddy model simulations from 1972 to 1990. Hake biomass is from a SST/biomass relationship derived by Ware and McFarlane (1994); hake catch is from Leaman (1992); adult herring biomass is from Schweigert et al. (1992); juvenile herring biomass is calculated from adult herring biomass (see text). All biomasses and catch are in thousands of tonnes. Hake Herring Year 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 Biomass 148 113 100 100 100 100 378 219 184 281 104 343 100 175 210 210 202 255 313 Catch 40 15 17 16 06 05 06 12 18 24 32 41 42 25 56 74 90 98 73 Adults 41 70 84 54 47 35 24 30 28 23 14 14 27 41 31 41 40 41 27 Juveniles 76 126 150 95 75 55 35 44 41 35 22 23 46 66 49 65 62 57 35 115 effects of different factors on these annual production estimates, before comparing model output to empirical data. I also compare simulated plankton and fish production with observed interannual variability in ocean climate. 3.3.1 Interannual simulations The annual diatom production generated from the standard 1972-90 simulation run ranges 250-500 g C m"2 y"1, with an average of 345 g C m ' 2 y_1 (SD= 60.7; Figure 3.1). Annual copepod production ranges 5-50 g C m"2 y"1 (mean= 16.4, SD= 11.3; Figure 3.2), while euphausiid production ranges 7-20 g C m y"1 (mean= 13.9, SD= 4.3; Figure 3.3). Characteristic of all three plankton time series is the high interannual variability in production, especially before 1983. The coefficient of variation (standard deviation/mean * 100) over the whole time series is 18% for diatoms, 69% for copepods, and 31% for euphausiids. Smoothing the 19 y plankton time series with a LOWESS smoother (Wilkinson 1990; tension = 0.5) reveals that production of all three groups increased from the early 1970s until mid 1980s, before declining during the late 1980s (Figures 3.1,3.2,3.3). Annual adult herring production ranges 0.6-2.6g C m"2 y"1 (mean= 1.14, SD= 0.58; Figure 3.4), while hake production ranges 0.5-1.9g C m"2 y"1 (mean= 1.23, SD= 0.43; Figure 3.4). Both adult herring and hake production time series show similar interannual variability when compared with the plankton (CV = 51% and 35%, respectively; Figure 3.4). Smoothing the fish production time series indicates that herring production declines from about 2.5 g C m"2 y"1 in the early 1970s to a low of about 0.5 g C m"2 y"1 by the early 1980s, before slowly increasing to 1 g C m"2 y"1 in the late 1980s (Figure 3.4). Simulated annual hake production shows the opposite trend compared to herring by increasing from the 1970s to mid 1980s, before declining in the late 1980s (Figure 3.4). Over the 19 y time series, both simulated annual herring and hake production are positively correlated with euphausiid production (r= +0.11 and r= +0.93, respectively), and negatively correlated with copepod production (r= -0.17and r= -0.65,respectively). 116 500 VV A * Afe & & >& 500 450 400 - 350 300 250 200 \ * Year Figure 3.1. Simulated estimates of annual diatom production for the Juan de Fuca Eddy from 1972 to 1990. The range in production in a given year (bars) is determined by using variable starting plankton concentrations for 1972 (see text). The long-term trend (dashed line) is determined using a LOWESS smoother (tension = 0.5). Year Figure 3.2. Simulated estimate of annual copepod production for the Juan de Fuca Eddy from 1972 to 1990. See Figure 3.1 legend. Year Figure 3.3. Simulated estimates of annual euphausiid production for the Juan de Fuca Eddy from 1972 to 1990. See Figure 3.1 legend. 5.0 CM £4. o 3 o o 3 S2 o -I" X 0.0 -\ H P \ \ 1 / \ NV » \ ^ 1* L **\ 1 . I . I i -r / i A. / Vs. i i i / I 1 HAKE • —| — /—\— a af \ \ 1 ^ >M^ "•"• HERRING y \ ^ i i i i i i i V — — i i i -— \ 2.5 2.0 c7~ E O O) 1.5 —' C o u o o 0.5 j * (0 /l*V A* Afe A* <£ ^ ft* <?£ <# ft0 v<y N v N<y N<v ^ N<p N*» N<p N o * N<s» o.o Year Figure 3.4. Simulated estimates of annual herring and hake production for the Juan de Fuca Eddy from 1972 to 1990. The long-term trends (dashed lines) were determined using a LOWESS smoother (tension = 0.5). Fourier analysis was conducted on the plankton annual production time series because it reveals the relative contributions of different "frequency waveforms" to the overall shape of the time series (Wilkinson 1990). The analysis indicates that each of the simulated plankton time series has a large spike at a period of 1 year. A similar result is obtained when the cycles in the upwelling season Ekman transport and start of the spring transition are analyzed (See below). Autocorrelation analysis of the differenced (i.e, trend removed) production time series confirms that copepods, euphausiids, and hake have 1 year lagged and significant, negative autocorrelations ( -0 .66, -0 .76, -0 .71, respectively). 3.3.2 Effect of factors on 1972-90 production estimates The short and long term 1972-90 standard run production patterns may potentially be influenced by a variety of factors; I evaluated four: 1) starting plankton concentrations used in 1972, 2) ending concentrations of one year on the next year, 3) perturbations in important zooplankton parameters, and 4) environmental forcing function variability. Starting concentrations: To evaluate the effects of starting (1972) plankton concentrations on subsequent years, I ran two simulations using the estimated starting range of plankton concentrations discussed in Section 2.3.1 (see Table 2.7). I assume that winters with low diatom concentrations have low zooplankton concentrations, while winters with high diatom concentrations have high zooplankton. This implies that plankton interactions are tightly coupled during the winter (Evans and Parslow 1985; Parsons and Lalli 1988). Only 3 of 19 years (1976, 1983, and 1990) have annual diatom production with > plus or minus 10% of standard run values (solid bars in Figure 3.1). Annual zooplankton production estimates are also typically within narrow ranges of standard run values (solid bars in Figures 3.2 and 3.3). These results indicate that the simulated plankton production patterns are relatively independent of the 1972 starting plankton concentrations. One year versus continuous simulations: To determine if conditions occurring in one model year influenced the remaining time series production dynamics, I compared results from the standard run continuous simulation to results from running each model separately. The starting conditions used for each model year are the same as those in the 1972 continuous simulation. The results of the continuous versus single run model experiment indicate that in the majority of years the absolute value of, and the change in direction of, annual plankton production is similar. For example, only at the end of the diatom time series is there a deviation in the direction of change from 1988 to 1989, and 1989 to 1990 annual production (Figure 3.5). The results of this model experiment indicate that conditions in the current model year generally do not influence the production estimates in the subsequent or remaining years. Parameter perturbations: Results from the sensitivity analyses in Chapter 2 indicate that several parameters might significantly effect absolute annual zooplankton production (See Table 2.8). The purpose of this section is to determine if changes in important zooplankton parameters, significantly effect the interannual production patterns. The following experiments are conducted: 1) euphausiid import is set to zero, 2) copepod offshore transport rate is reduced from a maximum daily rate of 13% d"1 to 10% d"1 (maximum and mean estimates, respectively from Mackas 1992), and 3) availability of copepods to euphausiids is reduced to 20% from 40%. Results from continuous simulations (1972 to 1990) with each of the three above parameters individually perturbed, indicate that copepod and euphausiid annual estimates range on average 1.5-1.65 times and 0.6-0.85 times standard run production, respectively (e.g., Figures 3.6,3.7). Diatoms and herring are less variable, ranging from 0.95-1 times and 0.9-0.97 times standard run values, respectively (Figure 3.6,3.7). These results indicate that plankton and fish sensitivity to zooplankton parameter variability is similar to the 1985-89 averaged simulation. Moreover, irrespective of the parameter being perturbed the Diatoms 122 500 «475 <N 450 O 425 2400 C .2 375 3 350 g 325 a _ 300 5 275 .5 Q 250 -225 ---— — • -/t \ 1 1 \ i h 1 V 1 1 \\ / I 1 ^ // // 1 1 year V i O \ / / \ \ / 1 1 1 ** -J t A / \ Continuous i i 1 ! --n j \ \ M n \K r _ ! 1 1 \* A* A* J> & A<J> N«" N* SP rf? N ^ N ^ o?> N* # «rf>~ 500 475 450 425 400 375 350 325 300 275 250 225 «?" Year Copepods 60 N£ 5 0 o 3 4 0 O 30^h •a o i -Q.20 TJ O a Q. O O \* 0 --h \\ r fll i i f 'A i \ . \J M [ i l l -Continuous ,1' -k A v A^ V -i i i i i i i i i 60 50 40 30 20 10 A* <& <& J§* <& «S>* <& <& n»° N<y N<»N fr ^ »°r £r N«° sqr N<s° Year Figure 3.5. Comparison of diatom and copepod production from the standard continuous simulation (solid lines) to production generated from running each model year separately (dashed lines). Diatoms 123 550 ~-500 CM £ U 450 § 4 0 0 1 350 "300 E o « 250 Q 200 — No — Lov\ _ a / \ / \\ i import / transport • l \ 7 \ // \ 7 y* i i i «. . r~~ ' / ^ . ' / / i i \Vi \ 1 \ \" ' \ 1 r \ L *\ m nA A , VY \ /' v\ \ 1 \\ jl * Jl 1 1 1 1 ~ I jl r V/ _ i o?> /i*V /> sP A * <?P <#• <bN 9$» <8" <8" N # N # s # N # N** ^ »JP ^ N<£* N<5» Year 550 500 450 400 350 300 250 200 Copepods 60 "No import " Low transport 60 Year Figure 3.6. Comparison of standard run diatom and copepod annual production (solid lines) to simulations without euphausiid import (dots), and to simulations with lowered copepod offshore transport (dashed lines). Euphausiids 124 \ * ' & ^ K# J ^ J> S Year 3.0 >» N 2.5 E O TO 2.0 C o 3 | , c Herring 3 til / r ' ~ I. • I Herring A A • \ * A 1 1 1 ! 1 1 i W i i — No — Low / T V / 1 1 1 import / transport -/*^V 1 I 1 3.0 2.5 2.0 1.5 1.0 0.5 ^ ^ ^ ^ f f f f f f 0.0 Year Figure 3.7. Comparison of standard run euphausiid and herring annual production (solid lines) to simulations without euphausiid import (dots), and to simulations with lowered coepepod offshore transport (dashed lines). 125 relative interannual and longer-term patterns in plankton and herring production are relatively similar to the standard 1972-90 simulation (Figure 3.6, 3.7). Environmental forcing function variability: To test the relative importance of the environmental forcing functions on the interannual patterns in plankton and fish production, I replaced a year's pattern in Ekman transport, or sunshine, and or SST, with the mean 1972-90 abiotic pattern. While one abiotic pattern is held constant among years, the remaining two abiotic factors retained their original seasonal variability. Results from the simulations indicate that all plankton production patterns are highly correlated with the standard run production patterns, except for those generated using mean 1972-90 Ekman transport (Table 3.2). Compare the standard run diatom and copepod production patterns to those interannual patterns generated using the mean 1972-90 Ekman transport function (Figure 3.8). Note that the mean Ekman pattern results in substantially lower time series variability, and different long-term production trends. Overall, the seasonal pattern in Ekman transport (via nutrient input) has the greatest influence on the interannual patterns in plankton and fish production. 3.3.3 Comparison of model output to empirical data The simulated interannual plankton production patterns should ideally be compared to or corroborated against empirical data. Long time series of plankton biomass or production do not exist for the southern BC shelf. Mackas (1994) however, has estimated anomalies in zooplankton biomass for the Eddy region (and the other two oceanic subregions) for 1985 to 1992. Correlations between observed biomass anomalies (from 1979-92 observed data) and simulated biomass anomalies (from the 1979-90 simulated mean) are low and nonsignificant. The trend in increasing euphausiid production at the end of the simulated time series is however, consistent with the results of Mackas (1994). The relatively weak relationships between the simulated and observed zooplankton data may be a result of a lack of interannual "contrast" during 1985-126 Table 3.2. Correlations between standard run plankton production and plankton production produced using mean 1972-90 Ekman transport, or sunshine, or sea surface temperature, or hake biomass patterns. The higher the correlation the more similar the interannual plankton production pattern is to the standard run patterns (See Figures 3.1, 3.2, 3.3). Note that using a mean Ekman pattern produces the production patterns with the least similarity to standard run patterns. For N=19, the correlation coefficient has to exceed ± 0.575 (**) or ± 0.456 (*) to be significant at the 1% and 5% level, respectively. Mean 1972-90 patterns Ekman Sunshine SST hake biomass No import Standard diatoms 0.31 0.60** 0.89** 0.93** 0.87** Standard copepods -0.20 0.93** 0.96** 0.95** 0.92** Standard euphausiids 0.21 0.82 0.86 0.80 0.94 Diatoms 500 "Standard run ' Mean Ekman 500 450 400 350 300 250 200 & «*&' $r v<y ^S" ff „o>v ^ fp ^ Copepods 50 . £ 4 0 o £ 30 o 3 TJ O 20 O Q-10 © a o O -\ m V V I I I w i M 1A / ! 1 1 1 1 Standard run "•"Mean Ekman -1 1 1 t 1 1 1 1 50 40 30 20 10 ^ /* /" ** f J J /> ^ ^ Year Figure 3.8. Comparison of annual diatom and copepod production from the standard run (solid lines) to production from a continuous simulation using the 1972-90 mean Ekman transport pattern (dashed lines). 128 90 (e.g., Figure 3.6). In addition, the expected small-scale temporal and spatial variability associated with sampling plankton roughly monthly, makes exact comparison with weekly averaged simulation results dubious. Intuitively, it is more reasonable to compare simulated output to temporally integrated measures of the system. To further test the efficacy of model output I evaluated the relationships between empirical data describing the condition of Pacific herring from 1972 to 1989 (D.M.Ware, DFO, unpublished data) and 1) simulated adult herring properties, and 2) simulated seasonal zooplankton production. Herring condition factor is a convenient way of comparing the relative interannual physiological state of age 5 herring. The larger the factor the heavier a herring for a given length, and presumably the greater its growth for that year. Observed condition of age 5 herring has increased roughly linearly and by about 30% from 1972 to 1989 (Figure 3.9). The condition factor is positively correlated with simulated herring growth (r= +0.10), and negatively correlated with starting model herring biomass (r= -0.19),and annual production (r= -0.30). Cross-correlation functional analysis of observed herring condition factor and simulated seasonal zooplankton production time series was conducted. Production by both copepods and euphausiids, copepods only, and euphausiids only in the spring (March, April, May), summer (June, July, August), and autumn (September, October) is determined for each model year. There are three significant positive, zero-lag correlations between herring condition factor and autumn zooplankton (r= +0.58; Figure 3.9), spring copepod (r= +0.50), and spring zooplankton production (r= +0.48). No significant time-lagged correlations were detected. 3.3.4 Relationships between production and ocean climate Correlation analyses between simulated zooplankton production and seasonal abiotic anomalies were conducted (Table 3.3). Seasonal abiotic anomalies (spring, summer, 129 Year 1.4 •_ 0 * • » o <" ^ o <— 1.3 C 0 "^ '•£ § 1 - 2 o O) c "C . . te 1-1 (V £ 0) Si-°i CO .o o A O ~~ — I • • B • • I • • • • • • • • I • • • 1 1 2.7 4.7 6.7 8.7 Simulated fall zooplankton production (gC/m ) Figure 3.9. Panel A: Condition of age 5 herring from 1972 to 1989. Panel B: Relationship between herring condition and simulated fall zooplankton production (r= +.58). 130 Table 3.3. Correlations among seasonal plankton production and corresponding seasonal abiotic anomalies (Sprg = March, April, May; Sum = June, July, August; Fall = September, October; ET = Ekman transport, SST= sea surface temperature, BS= bright sunshine). Note that lagged correlations between the current season's plankton production and the previous season's abiotic anomalies are included. For N=19, the correlation coefficient has to exceed ± 0.575 (**) and ± 0.456 (*) to be significant at the 1% and 5% level, respectively. Abiotic Anomaly ET SST BS Sprg Sum Fall Sprg Sum Fall Sprg Sum Fall Diatom Production Sprg +.52** +.35 +.29 - - -Sum +.12 +.49* -.27 -.17 -.22 +.28 Fall +.60** +.11 +.39 +.13 +.19 -.21 Copepod Production Sprg -.29 +.42 +.33 Sum -.43 -.51* -.24 -.12 -.19 -.19 Fall -.22 +.67** +.35 -.26 +.09 +.18 Euphausiid Production Sprg +.16 -.14 -.06 Sum +.71** +.29 +.09 -.06 +.27 +.04 Fall +.71** -.21 -.15 -.06 +.16 -.23 131 autumn, and winter) are calculated by subtracting the 24 y (1967-90) weekly average from each year's weekly value of SST and ET; BS anomalies are calculated by subtracting the 1972-90 weekly average. Diatom and copepod production is usually correlated with the current seasons abiotic conditions (e.g., spring ET and spring diatom production, r= +0.52; Table 3.3). Euphausiid production however, is more highly correlated with the past seasons abiotic conditions (e.g., spring ET and summer euphausiid production, r= +0.71). Overall, plankton production is most highly correlated with anomalies in spring and summer Ekman transport (Table 3.3). Compare the interannual and smoothed patterns in plankton production (Figure 3.1,3.2,3.3) with the patterns in spring and summer Ekman transport anomalies (Figure 3.10). To further characterize the ocean climate for each year, I clustered years with similar anomalous abiotic conditions during the upwelling season, using the Euclidean distance measure and the average linkage technique (Figure 3.11; Wilkinson 1990). Each group of years with similar ocean climate has characteristic zooplankton production. Group I contains years with moderately positive ET and BS, and moderately negative SST anomalies and has the lowest annual average diatom production and moderate copepod and euphausiid production (Table 3.4, Figure 3.11). Group II years are characterized by high positive ET, moderately positive SST, and moderately negative BS anomalies, and by high euphausiid, and moderate diatom and copepod annual production. The third group has years with very high positive anomalies in ET and BS, with low variable SST anomalies. Diatom and euphausiid production is highest in this group, with moderate copepod production (Table 3.4). The fourth ocean climate group has relatively high positive BS and SST, and variable ET anomalies. This group has moderate diatom, low euphausiid, and high copepod annual production. The last ocean climate group has the same plankton production trends as group IV but the two years are characterized by moderately negative ET and BS, and variable SST anomalies (Figure 3.11, Table 3.4). > A«> A * o?> ofr oJ> oS> o?> oS> A " -V* w ; OL» Ok* O ^ tt* S>" rfT rff <&> N<V & & & N<S" N< " N<S" N « " N<S" N « " Year B 140 Sea level Ekman transport 4 0 * _ J — I — L i i i i i i i i J ' ' '40 140 120 100 / ^ ^ S ^ / / ^ . / , / / / ^ Year Figure 3.10. Panel A: Interannual pattern in spring (dashed line) versus summer (solid line) Ekman transport anomalies (from 1972-90 mean). Panel B: Interannual variability in the day of the spring transition using Ekman data from 48° N compared to sea level data from 42°N (data from Hollowed 1990). -I g I I I I I I I I I I I I I I I I I I I I I I I I I .. t -Year Figure 3.11. Grouping of years by similiar abiotic anomalies measured during the upwelling season for 1972-90. Groups determined using the Euclidean distance measure and average linkage clustering technique (Wilkinson 1990). Numbers above groups refer to those listed in Table 3.4. 134 Table 3.4. Classification of years by anomalies in Ekman transport (ET), sea surface temperature (SST), bright sunshine (BS) (where + is positive, - is negative, L is low, M is moderate, H is high, and +/- is variable; refer to Figure 3.11). GROUP I II III IV YEARS 72,73,74, 75 77,80,84, 86,89,90 82,85,89 78,79,83, 88 Abiotic anomalies ET +M +H +H +/-SST -M +M +/-+H BS +M -M +H +H DIAT L M H M Annual prod COP M L/M M H uction EUPH M H H L 76,81 -M +/- -H M H 135 3.4 Discussion I have used the trophodynamics model and relationships with oceanic factors described in Chapters 1 and 2, to hindcast plankton and fish production on the southern BC continental shelf from 1972 to 1990. I start this section with a brief overview of the model processes that influence the production estimates and trends. I then discuss the short and long-term trends in plankton production in relation to oceanic variability. Finally, I discuss the temporal links and transfer efficiencies between plankton and fish. 3.4.1 Model limitations on production estimates It is necessary to be cognizant of model simplifications that may influence the hindcasted estimates of, and trends in, plankton production. Simulated diatom production, for example, may be overestimated in some model years because the negative influence of nitrogen depleted sub-surface waters on diatom growth is ignored. The model presently assumes that sub-surface nitrogen is constant among years. However, anomalously low nitrogen can occur along the west coast of North America, down to 200 m, during years of strong El Nino-Southern Oscillation conditions (Simpson 1992). Thus although upwelling favourable winds persist a substantially deeper thermocline results in warm, nutrient poor waters being transported to the surface layer (Simpson 1992), and ultimately in lowered diatom growth (Legendre 1990). How frequently does anomalously low nitrogen sub-surface water occur over the southern BC continental shelf? Brainard and McLain (1985, in Thomson and Ware 1988) noted that from 1951 to 1984 anomalously warm sub-surface water (to 100 m) occurred on the shelf as far north as coastal BC only in 1983. Since significantly warm and presumably nitrogen depleted sub-surface water occurs relatively infrequently over the southern BC shelf, simulated diatom production estimates are probably relatively unaffected by this process in the majority of years considered. The annual diatom production estimates would ultimately be enhanced with an improved understanding of the physical dynamics of nutrient supply to the mixed layer on the southern BC shelf. Certain model simplifications may also result in simulated annual copepod production being underestimated in some years (e.g., 1977) because one or two weeks of intense Ekman transport in early spring flushed already seasonally low copepod biomass out of the Eddy region. The resulting low copepod biomass generally did not recover until later in the autumn. The inclusion of a generalized offshore transport process in the model is well supported by observations made in coastal upwelling regions. Both Mackas (1992) and Verheye et al. (1992) indicate that high seasonal washout of copepods from shelf areas of upwelling regions is primarily due to seaward and alongshore transport. However, in nature, mechanisms likely exist for re-introducing copepod biomass to near-shore coastal regions effected by seaward transport. Verheye et al. (1992) propose an elaborate interaction between ontogenetic vertical migration in neritic copepods and subsurface currents to re-seed and maintain populations in coastal upwelling regions of the Southern Benguela. Inclusion of such a process in the present model would be desirable, but would require substantial data describing interactions between copepod population dynamics and sub-surface current structure. Annual euphausiid production in the model is strongly related to sub-surface import of euphausiid biomass from surrounding oceanic regions. This process was hypothesised to occur in the Juan de Fuca Eddy region to balance intense mid-summer predation by pelagic fishes (See Chapters 1 and 2). In the simulations, euphausiid import not only greatly enhances euphausiid production but also significantly reduces copepod annual production. Two important results are evident in comparing simulation runs with and without euphausiid import (Refer to Figure 3.6). The relative interannual pattern of annual copepod (or euphausiid) production remains similar, and the significant negative correlation between copepod and euphausiid annual production still exists (r = -0 .59 ;P< 0.05). The negative correlation is primarily due to explicitly modelling euphausiid predation on copepods. Is this pathway well founded? It is well documented that euphausiids feed on copepods in upwelling regions (Stuart and Pillar 1990 ), and that they can reduce copepod biomass (Verheye et al. 1992). Roff et al. (1988) also found a significant, negative correlation between euphausiid and copepod biomass. The authors deduced that euphausiids primarily effected copepod biomass during winter months, and this had consequences the following summer. The results of these studies combined with this study suggest that macrozooplankton can, at times, significantly impact mesozooplankton populations in coastal regions. The interesting implication is that macrozooplankton may compete with coastal larval fishes for spring and early summer copepod biomass. The dynamics of this competitive interaction may ultimately be related to the initiation and intensity of coastal upwelling, and thus the abundance of more "suitable" prey for euphausiids (e.g., diatoms; see Stuart and Pillar 1990). An evaluation of the simulation results indicates that parameter variability greatly influences absolute zooplankton production, but has little effect on diatom or herring production in a given year (See Chapter 4). This is because the parameters perturbed, directly effect zooplankton biomass. For instance, reducing the copepod offshore transport rate or eliminating the euphausiid import have relatively large positive effects on copepods and negative effects on euphausiid production. Perhaps the more important result is that while absolute values of production depend upon parameter values, the relative interannual patterns in production do not. This highlights the idea that one year's plankton production dynamics has relatively little impact on the next year's production dynamics. Several marine studies have alternatively suggested that biomass dynamics in winter significantly effect plankton production dynamics the following spring or summer (Colebrook 1984; Evans and Parslow 1985; Roff et al. 1988). These studies assume that biotic interactions strongly influence plankton biomass dynamics. However, it is likely that biotic processes are overridden by abiotic factors in more dynamic (variable) oceanic environments. Changes in the upwelling pattern, for example, results in large changes in 138 simulated diatom growth, and subsequently zooplankton and fish production. These results emphasize the relative importance of physical oceanographic variability in overriding biotic factors, and determining production in dynamic coastal systems. Bailey and Francis (1986) also indicate that it is likely that variable coastal environments are permanently under non-equilibrium conditions, and thus biotic factors will be secondary as regulating factors (See Chapter 4). I now discuss the short-term and long-term relationships detected between simulated plankton production and the variable ocean climate. 3.4.2 Interannual trends in plankton production One of the main results of this Chapter is the large interannual variability in plankton production (See Figure 3.1, 3.2, 3.3). These simulation results are consistent with Mackas (1994) who determined 3 to 10 fold interannual variability in copepod and euphausiid biomass anomalies for the period 1985-92 on the southern BC shelf. Large interannual changes in zooplankton biomass have also been detected further south in upwelling regions along Oregon and Washington (Peterson and Miller 1975; Mullin and Conversi 1988). Interannual variability in simulated coastal plankton biomass is also consistent with observations of high-frequency variability in spring copepod biomass measured offshore at Ocean Station P (see Fig 10 in Beamish and Bouillon 1993). One would expect a priori that the seasonally averaged copepod biomass in the two oceanic regions would be negatively correlated in a given year because of differing phytoplankton production responses to upwelling and associated atmospheric forcing (Figure 3.12). In the subarctic Pacific an intensified Aleutian low enhances mid-oceanic upwelling of limiting nutrients, thus phytoplankton growth, and ultimately copepod production (Venrick et al. 1987; Brodeur and Ware 1992). However, an intensified Aleutian low also results in increased poleward winds along the eastern Pacific and thus delayed and reduced coastal upwelling (Strub and James 1988; Simpson 1992), and ultimately reduced plankton Ocean Station P copepods (#/m ) Figure 3.12. Relationship between simulated spring (March-May) copepod biomass and that measured at Ocean Weather Station P during 1967-80 (data from McFarlane and Beamish 1993; r= -.34). 140 production (this study). The potential contrasting copepod production dynamics of coastal versus oceanic areas has important growth and survival implications for migratory fishes utilizing both regions (e.g., Pacific salmon). Most studies evaluating temporal trends in oceanic zooplankton, as influenced by ocean climate variability, focus on the mesozooplankton (e.g., copepods; see above). This is primarily due to the fact that accurate sampling of macrozooplankton, like euphausiids, is difficult because of patchiness and sampler avoidance (e.g., Simard and Mackas 1989). Results from this study however, indicate that euphausiid annual production or mean biomass may also vary significantly interannually, and at longer time scales (Figure 3.3). Simulated euphausiid production for example, is highest in years with above average upwelling (e.g., 1982). Under these conditions, biomasses of competitors for diatoms (copepods) and predators are lower (e.g., hake; see Ware and McFarlane 1994), while import of euphausiids from surrounding regions and prey (diatom) production is higher. Conditions are thus favourable for increased euphausiid growth (production). Brodeur and Pearcy (1992) noted an increased occurrence of euphausiids in the diets of pelagic fishes along the Washington/Oregon coast in years of greater upwelling, like 1982. Harris et al. (1992) similarly found that euphausiid biomass is roughly 3 orders of magnitude higher in coastal regions of Tasmania in windier years than in calmer years. These observations suggest that euphausiid biomass (or production) is linked to enhanced upwelling. Evidence of interannual variability in copepod and euphausiid production or biomass focuses attention on the possibility that the two zooplankton groups may be in or out of phase with each other in a given year (compare Figures 3.2 and 3.3). Note that I am not suggesting that zooplankton production along the BC coast is solely dominated by neritic copepods or euphausiids or combinations therein. There are certainly instances when other zooplankters (e.g., salps or oceanic copepods) may dominate zooplankton biomass and presumably production (Mackas 1992; Mackas 1994). The main point is that 141 there may be large variability in the production of dominant zooplankton groups relative to each other, as illustrated by model results of the present-day system (see Table 3.3). Roff et al. (1988) observed annual copepod and euphausiid biomass in the North Sea to be negatively correlated from 1974 to 1980. Likewise, Mullin and Conversi (1988) indicated that the median ratio of euphausiid to copepod biomass sampled at inshore stations over a submarine canyon off Monterey California, is 4 times higher during the non-El Nino years of 1955-57 compared to El Nino years of 1958-59; the change in the ratio is due to a decrease in euphausiid biomass during 1958-59. Harris et al. (1992) also indicate that krill (and salps) dominate zooplankton biomass in average windy years in the Tasman Sea, while copepods dominate biomass primarily during calm years. It may be insightful to develop interannual measures of variability in copepod versus euphausiid properties (e.g., mean biomass ratios; see Mullin and Conversi 1988) over time to examine integrated effects of upwelling on the different zooplankton groups. It is expected that the copepod to euphausiid biomass ratio would be primarily influenced by relative changes in euphausiid biomass in response to upwelling (Mullin and Conversii 1988; Harris et al. 1992). This expectation is consistent with the size selective feeding behaviour of zooplankton in coastal upwelling regions (see Introduction). The empirical and simulated evidence discussed above, concerning short-term variability in plankton production, seems convincing. Several studies have alluded to the idea that oceanic variability influences this high frequency variability in the plankton. In the simulations, the high frequency plankton production variability in the 1970s is clearly linked to the variability in the timing of the spring transition, and to the relative mean intensity of upwelling (Figure 3.10). Striking in this pattern is the biennial oscillation between early and late spring transitions, and the negative correlation with strong versus weak upwelling (Figure 3.10). After perusing the literature, uncertainty still exists as to the atmospheric mechanisms or dynamics generating this biennial phenomenon. Longer time series are probably required to determine whether the biennial pattern observed 142 during the 1970s is the norm, or whether the less variable pattern detected in the 1980s is. From these data though, it is becoming more lucid as to why simple linear relationships between fish properties like recruitment, and oceanic properties frequently fail. 3.4.3 Long-term trends in plankton production Another major result of this Chapter is the relatively long-term trend in simulated plankton production over the 19 y time series, superimposed over the high frequency variability. Several empirical studies have identified long-term trends in plankton biomass and also have linked them to variability in ocean climate. Dickson et al. (1988) for example, linked a long-term decline in copepod biomass in the North Atlantic, from the 1950s to 1970s, to the delay and reduced development rate of the spring phytoplankton bloom. Spring bloom dynamics were in turn related to the increased intensity in the Azores-Bermuda High pressure system. In contrast, Brodeur and Ware (1992) noted an increase in summer zooplankton biomass in the subarctic Pacific during the 1980s relative to the late 1950s, and related it to increased vertical mixing (nutrient input) in winter. Changes in mixing were related to enhanced winds associated with an intensified Aleutian Low. The long-term trends in simulated plankton production are similarly determined by the interactions between up welling and the dynamics of two large-scale atmospheric pressure systems, namely the Aleutian low and the North Pacific High. The initiation and intensity of coastal upwelling in spring, for example, is negatively influenced by a strong Aleutian Low because of increased southeasterly winds along the coast (ALPI of Beamish and Bouillon 1993 versus spring ET or spring transition; r= -0.29 and r=-0.15, respectively). Reduced spring coastal upwelling is further corroborated by a positive correlation between ALPI and spring SST (r= +0.75; see also McFarlane and Beamish 1992). Summer coastal upwelling on the other hand, is influenced by the dynamics of the Pacific High and Continental low pressure systems (Strub and James 1988; Bakun 1990). The observed 143 long-term increasing trend in summer upwelling from the 1970s to mid 1980s (See Figure 3.10) is due to enhanced warming over land, and subsequent intensification of northwesterly winds (Bakun 1990). The atmosphere-coastal upwelling interactions effect simulated diatom production by influencing growth via nitrogen limitation and offshore export. Copepod production is effected via offshore export, prey production, and euphausiid predation, while euphausiid production is influenced by prey production, import from surrounding deep-water sources, and by predation from migratory hake. The important result is that trends in simulated annual plankton production are determined by the combined dynamics of spring and summer upwelling. For example, 1985 had positive spring and summer upwelling anomalies and high diatom production, while 1984 had negative spring and positive summer upwelling anomalies and average diatom production (see Figure 3.1 and 3.10). Although the combination of spring and summer upwelling is primarily responsible for determining simulated annual plankton production, trophodynamical phasing can modify the expected outcome. Because 1976 had negative spring and summer upwelling (Figure 3.10), it would be expected that plankton production be relatively low. However, extremely high diatom and copepod production is simulated (see Figures 3.1 and 3.2). Reduced upwelling of nitrogen in 1976 limited diatom growth, and thus copepods were better able to track diatoms. Increased trophodynamical phasing (sensu Parsons 1988) results in three very large oscillations in nitrogen, diatom, and ultimately copepod concentrations during the upwelling season. This enhanced limit-cycle behaviour may be due to the inherent mathematical properties of the functional type II response used to describe the interactions between nitrogen, diatoms, and copepods. However, characterization of plankton feeding interactions using a Type II response is well documented (see Parsons et al. 1984), and the 1976 results support the notion that copepod biomass should be higher in less windier years (see below; Harris et al. 1992). It is interesting to note that the anomalously high simulated copepod production in 1976 occurred immediately before what Hollowed and Wooster (1991) and Beamish (1992) have suggested is a year (1977) of synchronously strong year-classes of marine fish along the west coast of North America. 3.4.4 Fish Production The above sections focused on relationships between ocean climate and plankton production. In this section, I discuss the relationships between plankton and fish. The result that euphausiid production is temporally variable has important implications for adult pelagic fishes like Pacific hake and herring, because euphausiids are their main prey (Tanasichuk et al. 1991, Ware and McFarlane 1994). Potential growth variability in adult hake may be extremely important to stock dynamics given the fact that they undergo extensive return seasonal migrations from northern feeding grounds to spawning grounds off southern California (Ware and McFarlane 1989). Reduced prey (euphausiid) production on northern feeding grounds may ultimately negatively effect subsequent egg production or quality. The combination of reduced prey productivity on northern feeding grounds and on recruitment grounds off southern California was recently hypothesized as a major factor contributing to the demise of the migratory Pacific sardine (Sagax sardinops) in the 1940s (Ware and Thomson 1991). Variability in euphausiid biomass may also have profound effects on distributions of pelagic fishes, and thus the relative success of the fishery. Harris et al. (1992) observed that the Pacific mackerel fishery off Tasmania is more successful in years of greater upwelling because of increased biomass of fish in areas of enhanced euphausiid aggregations. In warmer years mackerel stocks "break-up" because large aggregations of euphausiids do not occur. The distribution of hake stocks on the BC shelf exhibit a similar response to changes in euphausiid biomass as determined from hydroacoustic surveys and as estimated by water temperature (Ware and McFarlane 1994). Results of this study identify two significant correlations between observed 145 herring condition and simulated zooplankton production. Herring condition is significantly and positively related to simulated spring copepod and spring zooplankton production. This may be a critical period for herring because adults return to the Eddy region from spawning in Barkley sound in March/April . High spring zooplankton production and biomass may be essential for re-building body mass lost to spawning. Copepods may be more important than euphausiids to post-spawning herring because of higher oil content, or higher spring concentrations. Wailes (1936) observed that BC coastal adult herring consumed copepods (e.g., Calanus) in spring, and euphausiids the remainder of the year. Incorporating this pathway into the model may provide new insights into interannual variability in herring growth. Annual herring production is also significantly correlated with autumn zooplankton production. This result is consistent with data of R. Tanasichuk (DFO, Nanaimo) that growth of La Perouse herring is about 2 times higher in the autumn than the summer. Part of this seasonal growth difference is due to the fact that in summer, herring remain in the less productive and peripheral regions of the Eddy to avoid predation by hake, and thus their growth is lowered (Ware and McFarlane 1994). In fact, the simulations indicate that there is a possible non-linear relationship between herring condition and euphausiid summer production (Figure 3.13). Thus, as summer euphausiid production increases, Pacific hake remain in the Eddy region to feed, and thereby reduce herring production. In studies of marine systems, ecological transfer efficiencies are typically used to estimate fish production from primary production (e.g., Ryther 1969). The commonly used equation is: FP = ACP * E n * c, where FP is fish production (g WW m"2 y"1), ACP is annual phytoplankton production (g C m y"1), E is the carbon transfer efficiency, n is the number of "trophic levels", and c is the carbon to wet weight conversion factor. It is frequently assumed that transfer efficiency (E) in the above equation is constant at 10%. This assumption is likely not valid because of spatial and temporal variability in transfer Simulated euphausiid production (g C/m ) Figure 3.13. Relationship between euphausiid summer production and observed herring age 5 condition factor for the period 1972-89 (unpublished data from D. Ware; r= -.33). 147 efficiencies due to the dynamics of feeding interactions, and because the magnitudes of non-trophodynamical losses are usually unknown (Parsons et al. 1984; Iverson 1990). The results of the trophodynamics model provide some additional insights into transfer efficiency variability. Over the time series, the annual diatom-to-fish transfer is about three times as variable as the diatom-to-zooplankton transfer (CV = 78% versus CV = 26%, respectively; Figure 3.14). The higher variability in the diatom-to-fish transfer is due to the large decline in herring biomass (production) from the early to late 1970s (See Figure 3.4 and Schweigert et al. 1992). Also note that both the diatom-to-zooplankton and diatom-to-fish transfer exhibit large changes (e.g., doublings) in consecutive years. The majority of the diatom-to-fish variability in any one year comes from the diatom-to-zooplankton component of the transfer. For example, the monthly diatom-to-euphausiid transfer from June to November, is about 1.3-1.6times more variable than the euphausiid-to-fish transfer. In light of this result, I now briefly discuss some of the important seasonal aspects of the diatom-to-zooplankton transfer. An evaluation of the diatom-to-zooplankton transfer over the year identified several important characteristics (Refer to Figure 3.15). The lowest TE occurs in winter because of relatively low overwintering zooplankton biomass and little or no diatom production due to light limitation. Diatom production increases rapidly in spring, but slower growing zooplankton result in relatively low transfer efficiencies (e.g., April in Figure 3.15). This temporal lag in zooplankton biomass growth is an inherent property of marine systems dominated by large phytoplankters and metazoans (Kiorboe 1993). The diatom-to-zooplankton TE from June to August is relatively constant (9-10%) because of close trophodynamical phasing between seasonally high zooplankton biomass and high diatom production (Figure 3.15). Diatom production is highest during this period because of the almost constant upwelling of nitrogen associated with the stability of the Pacific High pressure system in summer. The CV in summer upwelling, from 1972 to 1990, 148 16.0 Figure 3.14. The interannual pattern in the efficiency of transfer from diatoms to zooplankton (euphausiids and copepods) and from diatoms to fish (hake and adult herring). Transfer efficiency is calculated by dividing fish or zooplankton production by diatom production and multiplying by 100. Month Figure 3.15. Monthly diatom (D) to euphausiid and copepod (E + C) transfer efficiency and diatom production averaged over 1972-90. The bars represent the 95% confidence interval around the mean transfer efficiencies. 150 is about 5 times lower than the CV in spring or autumn upwelling (Figure 3.16). Interestingly, during the summer upwelling period mean zooplankton TE remains within a relatively constant range (9-10%) even though mean diatom production ranges from 65-100 g C m . Offshore transport ultimately limits the transfer because slow growing zooplankton are unable to replace their exported biomass as quickly as the faster growing diatoms can. The highest diatom-to-zooplankton transfers occur in the autumn (Figure 3.15) because of high zooplankton biomass (from summer) and reduced diatom growth. Diatom production is low in the autumn because of increased light limitation caused by an increased likelihood of intense southeasterly winds, and thus downwelling (Figure 3.16). The seasonal dynamics of the diatom-to-zooplankton transfer is a result of the influence of upwelling on diatom production and subsequent trophodynamical phasing with the zooplankton (See Parsons 1988; See Chapter 4). This result is analogous to the TE-turbulence relationship described among marine environments (e.g., Cushing 1971). 500 450 ^ 400 O 350 "C 300 O W 250 C <" ««« *. 200 j§ 150 E . * 100 ID 50 \ N \ ^ «* ^ J +* /> ^ jfi cf O* ^ j Date Figure 3.16. Coefficient of variation (CV) in Ekman transport averaged weekly from 1972 to 1990. 152 CHAPTER 4: INTERACTIONS BETWEEN RESOURCE LIMITATION AND PREDATION IN A COASTAL UPWELLING FOOD-WEB 4.1 Introduction The variability in "trophic level" biomass or production in aquatic systems, in general, is primarily due to interactions between resource and predator limitations. There are however, relatively few studies that specifically document interactions between resources and predator limitation in coastal marine systems. The paucity of studies is linked to confounding effects associated with spatial and temporal variability in horizontal and vertical mixing (e.g., tides and upwelling), to inadequate and infrequent sampling, and to seasonally local and regional ontogenic migrations of vertebrates and invertebrates (see papers in Nixon 1988). The evidence for resource limitation, from the few studies that have been conducted in marine systems, comes from considering: 1) the positive influence of nutrients like nitrate or iron on phytoplankton growth (see papers in Chisholm and Morel 1991), 2) the positive influence of phytoplankton concentrations on growth and reproduction of coastal copepod populations (e.g., Kiorboe 1991), 3) the match between spring copepod production or biomass and larval fish survival/growth (Cushing 1975; Wroblewski and Richman 1987; Lasker 1988), and 4) the positive relationship between primary and fish production (e.g., Ryther 1969; Cushing 1975; Iverson 1990). The influence of predator limitation on trophic level biomass in coastal marine systems is less well documented than for resource limitation. The majority of published studies consider the impact of both invertebrate and vertebrate marine predators on copepod populations (e.g., Koslow 1983; Ohman 1986), while only a few studies have documented the predatory impacts of coastal marine fish on other fish (Ware 1989), or fish on zooplankton (Rudstam et al. 1992). 153 Although relatively few studies have considered the interactions between, and effects of, resources and predators acting simultaneously on major trophic levels in coastal marine systems, the freshwater literature is replete with studies evaluating these processes (Carpenter et al. 1987; McQueen et al. 1989 and references therein). Two conceptual models have emerged from the plethora of recent studies in productive (eutrophic) lakes. One freshwater model generally predicts that changes in predator biomass at the top of the food-web (e.g., piscivores) will cause phytoplankton biomass to deviate from nutrient predicted concentrations (Carpenter et al. 1985, 1987). Specifically, the "trophic -cascade" model predicts that increased piscivore biomass causes a decline in planktivore biomass, which leads to increased zooplankton, and ultimately decreased phytoplankton biomass. One important characteristic of top-level predators in eutrophic lakes that influences trophic cascades, is their inertia. Predatory inertia results primarily from successive year-class failures or successes by piscivores (Carpenter et al. 1985). Note that in coastal marine systems predatory inertia is a more likely a combination of temporal variability in local fish recruitment, and of the biomass and body-size of migratory predators, like Pacific hake (See Chapter 2). A second general conceptual model from the freshwater literature extends the trophic-cascade model by predicting that maximum attainable biomass at any trophic level is determined by processes influencing nutrient availability (bottom-up processes, BU), but that actual biomass is determined by the combined effects of BU processes and predators (top-down processes, TD). The BU.TD model also predicts that TD forces are strong at the top of the food-web but that they dampen towards the bottom. Conversely, BU processes are strong near the bottom and weaken towards the top of the food-web. In productive aquatic systems then, trophic interactions should damp out as they cascade down the food-web such that phytoplankton biomass or production will show little response to variable top-level predator biomass (McQueen et al. 1986, 1989). Given the above discussion, it is apparent that there are at least two major 154 differences between eutrophic lakes and coastal marine systems that may influence the strength of TD effects on lower trophic level biomass. One striking difference is the presence of large-bodied omnivorous or predatory zooplankters in coastal marine systems (e.g., euphausiids or chaetognaths; Lehman 1988). With the increased complexity of intermediate trophic levels in coastal marine food-webs, it may be expected that a TD perturbation would be dispersed and thus dampened before reaching the phytoplankton (sensu Denman et al. 1989). A second major difference in coastal marine systems is the fact that most top-level pelagic fish predators are not strictly piscivorous. Most adult coastal marine fishes feed primarily on the relatively large biomasses and aggregations of macro zooplankton mentioned above (Tanasichuk et al. 1991; Brodeur and Pearcy 1992). The feeding habits of coastal fishes imply that they may have stronger TD impacts on primary producers, than do the strictly piscivorous fishes of productive lentic systems, simply because the former are only one "trophic level" away from the phytoplankton. To sum up, relatively few studies have evaluated the relative importance of TD versus BU effects in the more trophodynamically complex marine systems. The simulated "food-web" should enhance our understanding of the interplay between BU and TD processes because many confounding factors typical of marine systems have been simplified. The main objectives of this chapter are to use the simulation model of Chapter 1 and the 1972-90 simulation results in Chapter 3 to: 1) characterize the relative strengths of resource versus predator limitation on plankton and fish biomass and production, and 2) describe and quantify the seasonal and interannual variability in these processes. 4.2 Methods To fully understand factors effecting flows of biomass (or energy) through a system of organisms, one must consider their production dynamics (Parsons et al. 1984). Since estimates of production are more difficult to determine than biomass, most empirical 155 studies have focused on changes in the latter. However, the simulation model will provide insights into how plankton and fish production may be effected by variations in BU/TD processes. To assess the relative strengths of bottom-up (BU) versus top-down (TD) effects on model plankton and fish production I evaluated the correlations among annual plankton and fish production, and the trends in production over the range of Ekman transport and hake biomass used in the 1972-90 simulations (See Chapter 3). I have also conducted two model experiments. The first model experiment examined the TD effects of hake on plankton and herring by running eleven separate simulations (1 model year each). Each simulation used a different hake biomass (126 to 336 kt by 21 kt increments), while all other parameters are held constant. Recall that the results of the 1972-90 simulations are potentially influenced by export and import of plankton, as related to Ekman transport (see Chapter 2). Thus to remove the effect of Ekman transport, a second model experiment was conducted to examine the effects of nitrogen addition on plankton and fish by varying the upwelling rate from 2.4 to 6.4 m d"1 (by 0.4 m d"1 increments). In varying the upwelling rate, only the amount of nitrogen added to the mixed layer is changed. The relative rates of offshore plankton transport or euphausiid import remained constant. These simulations are analogous to varying deep-water nutrients (See Chapter 3). The range of nutrient input used had the same coefficient of variation as the hake biomass experiment (i.e., CV = 30%). For results of the model experiments I evaluated changes in coefficients of variation, general linear trends, and regression statistics. These characteristics are more descriptive or indicative of interactions among factors than simply evaluating means (Harris and Griffiths 1987). 4.3 Results and Discussion The trophodynamics model has 5 major "trophic levels" that can be influenced by 15 major BU/TD interactions (Figure 4.1). In the remainder of this Chapter I evaluate the HAKE 156 HERRING 8 EUPHAUSIIDS COPEPODS DIATOMS t NITROGEN Figure 4.1. Conceptual overview of the relative strength of "bottom-up" (odd #s) versus "top-down" (even #s) influences on trophic-level biomass as discussed in the text (thick solid lines: strong effect; thin solid line: moderate effect; thin dashed line: weak effect). 157 empirical and simulated evidence for the relative strengths of these BU versus TD interactions, and their influences on coastal plankton production. In section 4.3.1,1 discuss the BU interactions (labelled as odd #s in Figure 4.1), while in section 4.3.2 I discuss the TD interactions (labelled as even #s in Figure 4.1). In the last section, 4.3.3,1 discuss the possible temporal variability in the relative importance of BU versus TD interactions. 4.3.1 Bottom-up interactions The main a priori expectation of this study is that BU effects (i.e., resource limitation) primarily determine trophic level properties in coastal upwelling regions. This expectation is supported by empirical evidence that coastal marine fish production is positively correlated with annual phytoplankton production (Nixon 1988; Iverson 1990; Ware 1992). The 1972-90 simulation results also indicate that hake production is positively correlated to annual diatom production (r= +0.36; Table 4.1). However, empirically derived primary production-fish correlations frequently "break-down" over time (Rothschild 1991; Mann 1992). The lack of predictability or constancy in the expected response suggests that BU interactions are dynamic. In light of this, I evaluated each of the major interactions included in the model (Figure 4.1), to enhance our understanding of BU processes. Nitrogen-diatoms: In model simulations, variability in parameters or processes controlling nitrogen input to the mixed layer always have positive and significant effects on annual diatom production and mean annual biomass. For instance, in the upwelling rate model experiment and in the 1972-90 simulations, diatom production is positively correlated with nitrogen input (Figure 4.2). As expected, about 2 times more variance in diatom production is explained in the model experiment than the 1972-90 simulation, because additional effects of offshore transport, and other limiting factors like sunshine, 158 Table 4.1. Correlations between annual plankton and fish production generated from the 1972-90 simulation (See Chapter 3). For N = 19, the correlation coefficient has to exceed ± 0.575 (**) and ± 0.456 (*) to be significant at the 1% and 5% level, respectively. Diatoms Copepods Euphausiids Herring Hake Diatoms 1.0 +.09 +.28 -.19 +.36 Copepods 1.0 71** -.17 -.65** Euphausiids 1.0 +.11 +.93** Herring 1.0 -.20 Hake 1.0 159 Relative Upwelling rate B 500.0 g 400.0 O 3 c 300.0 O " Diatoms r Copepods "Euphausiids 50.0 Ekman transport anomaly Figure 4.2. Annual plankton production plotted against upwelling rate (Panel A) and against observed Ekman transport anomaly (Panel B). 160 are at work in the latter (Table 4.2). The positive correlation between diatoms and nitrogen is consistent with observations made by Probyn (1985) that net phytoplankton blooms develop rapidly in response to large inputs of nitrates into the euphotic zone via upwelling (see also Legendre 1990). Simulated and real world net phytoplankton (e.g., diatoms) respond rapidly and positively to enhanced nitrogen input to the mixed layer, and thus I have labelled BU interaction # 1 in Figure 4.1 as "strong". Plankton-plankton: Simulation results indicate that the BU effects of diatoms on zooplankton are strong (BU # 3 and BU # 5 , Figure 4.1). For instance, in the sensitivity analyses any parameter change effecting diatoms resulted in large changes in copepod seasonal biomass patterns (e.g., high loglikelihood f value in Table 2.8). The positive correlations between zooplankton and diatoms (r = +0.09 and +0.27, copepods and euphausiids, respectively) implies that zooplankton are potentially food-limited in this coastal upwelling system. Food-limitation in coastal copepods is a highly contentious issue in the marine zooplankton literature (e.g., Huntley and Boyd 1984). For upwelling regions however, it is generally agreed that zooplankton are seldom food-limited because phytoplankton respond 10-30 times more rapidly to favourable conditions than zooplankton, and because daily primary production rates are typically > zooplankton consumption (Mann and Lazier 1992). Zooplankton populations may lag numerically behind phytoplankton increases, but individual zooplankter growth is seldom limited by diatom production. The simulation results do however, indicate that copepod growth is prey-limited at times coincident with intervals between phytoplankton blooms (see also Walker and Peterson 1991). This is consistent with observations that diatom production in upwelling regions is analogous to a series of spring blooms (Mann and Lazier 1992). Occasional growth-limitation of model copepods may ultimately be influenced by simplifications assuming that diatoms are the only prey. In reality, copepods may be carnivorous under lower diatom concentrations (e.g., Landry 1981), or survive low food conditions on accumulated lipids (Wyatt 1980). 161 Table 4.2. Regression statistics for mean annual plankton and fish production plotted against gradients in hake biomass or Ekman transport anomalies for A) the 1972-90 simulations and B) the model experiments (See Figures 4.2 and 4.5; r: correlation coefficient; CV: coefficient of variation; b: regression slope) For N = 19, the correlation coefficient (r) has to exceed ± 0.575 (**) and ± 0.456 (*) to be significant at the 1% and 5% level, respectively. 1972-90 data Diatoms Copepods Euphausiid Herring Variable hake biomass b -0.057 +0.013 -0.007 -0.003 r -0.09 +0.10 -0.14 -0.44 2 r 0.01 0.01 0.02 0.20 Ekman transport anomaly b 8.15 -0.608 +0.389 -0.013 r +0.74 -0.40 +0.49 -0.11 2 r 0.55 0.16 0.24 0.01 Experiments: Diatoms Copepods Euphausiid Herring Hake Variable hake biomass b -1.05 +0.78 -0.29 -0.036 -0.002 r -0.35 +0.88** -0.98** -0.98** -0.83** 2 r 0.12 0.77 0.96 0.97 0.69 CV 7.0 11.0 8.0 8.0 3.6 Upwelling rate b 28.5 4.2 0.5 +0.009 +0.011 r 0.98** 0.97** 0.86** 0.83** 0.86** r2 0.96 0.95 0.73 0.69 0.74 CV 27.0 51.0 16.4 2.3 18.0 60 <M 50 o 3*o o 3 o '20 •o o a o a o O 10 250 • • • - A ~ A A • A * t A I • • - -* A * I • • A • • _ -• A A A A A I "*"Copepods "•"Euphausiids ~ - " "~ • A _ I 25 20 CM E O o> 30^ - -C 15 O *3 U 3 "O "I 3 (S £ a 3 lil 300 350 400 450 500 Diatom production (g C/m /y) Figure 4.3. Annual copepod (solid line) and euphausiid (dashed line) production plotted against diatom production. (r= +.09, and r = + .27, respectively). 163 Irrespective of whether or not copepod populations are occasionally prey-limited in upwelling systems, it is apparent that other factors, such as offshore export or invertebrate predation, cause copepods to deviate from expected BU effects. For example, model copepods are positively correlated with relative nitrogen input (without offshore transport; Figure 4.2a), but negatively correlated with observed 1972-90 Ekman transport anomalies (nitrogen input plus offshore transport; Figure 4.2b). These results support Mackas's (1992) hypothesis that the seasonal decline in copepod biomass on the southern BC shelf is due primarily to net offshore transport. Pitcher et al. (1992) also suggest that physical influences override biotic interactions during upwelling periods in the Benguela system. Furthermore, the latter authors indicate that when physical processes are less dynamic, biological interactions become the major determinants of plankton variability. It is probably more reasonable that larger zooplankton, like euphausiids, are primarily food-limited in coastal regions (Figure 4.2), because they have relatively high maintenance ration requirements, a greater dependence on larger prey (e.g., diatoms), and a tendency to form dense aggregations (Moloney 1992). It is important to note that BU effects of diatoms on euphausiids are mitigated primarily by the TD effects of Pacific hake (see below). Model euphausiids are potentially influenced by two BU processes, BU #5 from diatoms and BU # 7 from copepods (Figure 4.1). Since diatoms, on average contribute a larger fraction to model euphausiid production than copepods (e.g., 90% versus 10%; Figure 2.14), BU # 5 effect is stronger. Pillar et al. (1992) also indicate that E. lucens in the Benguela system is a preferential herbivore, with the bulk of the nocturnal diet being dependent upon the relative abundance of copepods and phytoplankton, and typically consisting of larger phytoplankton (See also Stuart and Pillar 1990). It is however interesting to note, that perturbations in model copepod parameters generate about 2 times greater variability in euphausiid production, than do perturbations in diatom parameters (Table 4.3). Processes reducing copepod biomass result in up to an 8% increase in 164 Table 4.3. Mean absolute percent deviation in annual plankton and fish production from standard run values (see Table 2.4), in response to perturbations in model parameters belonging to major groups discussed in Table 2.9. Interaction # s refer to Figure 4.1. Parameter group perturbed Diatom parameters Copepod parameters Euphausiid parameters Herring parameters Abiotic parameters State Variable Diatoms Copepods Euphausiids Herring Hake Diatoms Copepods Euphausiids Herring Hake Diatoms Copepods Euphausiids Herring Hake Diatoms Copepods Euphausiids Herring Hake Diatoms Copepods Euphausiids Herring Hake Mean % change 1.4 9.2 1.9 0.5 3.1 1.6 23.5 4.6 0.8 7.5 1.7 9.2 8.3 1.0 10.8 1.7 2.1 1.5 3.8 3.1 4.7 14.2 3.6 3.1 4.7 Standard deviation 1.9 5 1.3 0.45 2.8 1.2 15 2.8 0.7 5.1 1.4 10.6 5.1 0.6 7.3 2.0 1.8 0.8 1.6 2.8 4.5 9.7 2.1 6.9 3.2 Interaction # BU # 3 BU # 5 TD # 2 BU #7 BU # 9 TD # 4 TD # 6 BU # 9 BU #12 TD # 8 TD #10 BU #15 BU # 1 165 euphausiid production. This result implies that lower copepod grazing on diatoms increases the concentration of diatoms available to euphausiids. Empirical evidence indicates that E. lucens, in the southern Benguela system, "adjusts" its feeding behaviour in accordance with not only prey size and quality, but with the degree of interspecific competition within the zooplankton community (Pillar et al. 1992). Plankton-fish: Simulated copepods have only a weak BU effect on herring (BU #9) . In fact, sensitivity analyses indicate that perturbations in copepod parameters result in < +/- 2% changes in annual herring production (Table 4.3). It is possible that BU # 9 interaction is stronger in nature, because there is some evidence that adult herring feed on copepods in the spring (Wailes 1936). This pathway is not however included in the trophodynamics model. Copepods are expected to have a relatively moderate to strong BU impact on juvenile herring, but since the latter's production dynamics are not fully modelled, this interaction can not be adequately explored. The latter expectation is ultimately related to the importance of spring copepod production or biomass to growth and survival of juvenile pelagic fishes (e.g., Parsons and Kessler 1986; Wroblewski and Richman 1987). In fact, herring recruitment dynamics are thought to primarily influence stock biomass (rather than growth; D.M.Ware, DFO, Nanaimo, pers. comm). It is apparent that BU effects of zooplankton on herring need to be more fully evaluated. To accomplish this, data describing seasonal variability in both juvenile and adult herring diets, on the southern BC shelf, are required. Model euphausiids have a 10 times greater BU effect on hake (BU #13), than on adult herring production (BU #15; Table 4.3 and Figure 4.4). Euphausiids have a strong BU effect on hake because the former makes up a large fraction of the latter's diet (Figure 2.13). Fish-fish: Simulated herring have a weak BU effect on hake because they made up a small fraction of the hake diet (BU # 15). In addition, perturbations in adult herring parameters generate only a 3% absolute mean change in hake annual production (Table Euphausiid production (g C/m /y) Figure 4.4. Annual herring and hake production from 1972-90 simulation plotted against annual euphausiid production (r= + .11, and r= +.93, respectively). 4.3). It is possible however, that herring (and other small fishes) may have a greater BU impact on hake populations dominated by larger-sized individuals, because these bigger fish are more piscivorous (e.g., Livingston 1983). 4.3.2 Top-down interactions Fish-fish: At the top of the model food-web, hake have strong TD effects on simulated herring biomass (TD #14; Table 4.2). The strong TD impact of hake on herring has been hypothesized by Ware and Thomson (1986) as being mainly responsible for the negative correlation between water temperature and lower west coast of Vancouver Island herring biomass; recall the positive correlation between water temperature and hake biomass off southern BC (See Chapter 3). Oscillations in Hecate Strait herring recruitment have also been attributed to predatory (TD) impacts of Pacific cod (Walters et al. 1986). The TD impact of model hake on herring is ultimately influenced by the production dynamics of the euphausiids because the biomass of herring eaten by hake is a largely a function of the fraction of euphausiids in the hake diet (see Chapter 1; Tanasichuk et al. 1991; Ware and McFarlane 1994). Fish-plankton: Hake also have a strong TD effect on the next lowest fractional trophic level in the model, that occupied by the euphausiids (TD #12). The simulations indicate that the TD impact of hake on euphausiids is strong enough that a reduction in hake causes an increase in annual euphausiid production (Figure 4.5). The simulated TD impact of hake is strong because euphausiids constitute a large fraction of the hake diet (Figure 2.13), and because euphausiid biomass cannot outgrow hake predation (Figure 2.22). Results from the only empirical study I could find do not confirm the model prediction. Mullin and Conversii (1988) hypothesized that the initiation of the hake fishery along the Oregon/Washington coast in the mid 1960s would remove the euphausiids main predator, and thus euphausiid biomass or individual size should increase. The 168 --* 500.0 04 E W 400.0 3 c o o 3 o 1 -a E o (B w 3 C 300.0 " Diatoms "Copepods "Euphausiids _ A-- -A- "X 200.0- -! - - • - - • -100.0 0.0 _ — —A • -• -*• - -•-50.0 _ , 40.0 E O c o ,J I 30 .0 3 •o o 20.0 J c a a '10.0 8 « 3 C 0.0 Relative Hake biomass 600.0 CM 500.0 £ o O) 400.0, c n productio O O O O b b o * 1000 , 5 ' i i k 1—•— I B • A • • • A A A • • • "*• Diatoms "^Copepods "*" Euphausiids • • • • _ A • _ 60.0 50.0 40.0 30.0 E O 3 c o S3 u 3 •o o 20.0 10.0 0.0 c o S c a a. o o N c£' <&' cP' 4s ' cP' «£' cP* \° «J» <£» ($> n j 5 n? {P Hake biomass (k tonnes) Figure 4.5. Simulated annual plankton production plotted against relative (A) and 1972-90 estimated (B) hake biomass. authors could not detect significant changes in pre versus post hake fishery euphausiid biomass or body size. Mullin and Conversii (1988) indicate several possible reasons for not detecting the hypothesized TD impact: too few euphausiid biomass samples made primarily offshore and to the south of the hake fishery, a potential increase in alternative vertebrate euphausiid predators, and/or the euphausiids experience a switch from predator to food limitation. The model results indicate that sampling considerations are critical for detecting a TD impact of hake on euphausiids. Any increase in euphausiid biomass would be relatively short-lived because adult euphausiids not eaten by predators would die of senescence, and thus the "temporal-window" of detecting increased euphausiid biomass due to reduced TD impacts is likely small (See Figure 2.9). Of the two other possibilities, alternate predators or prey limitation, the latter is more likely to limit euphausiid populations, for two reasons. No extant marine vertebrate predator could quickly replace the approximately 60 kt (1980s average) of hake biomass removed annually, by the fishery. In addition, food-limitation may be an important regulatory mechanism in highly aggregated animals, like euphausiids (e.g., Pillar et al. 1990). Because of their large body-size and aggregations relative to copepods, euphausiids are ideal prey for size selective predators, and thus should exhibit the response predicted by the model (Simard and Mackas 1989; Pillar et al. 1992). Model herring have a moderate TD impact on annual euphausiid production, and a strong TD impact on copepods (TD #10 and TD #8 , respectively). Copepods are impacted directly via predation from juvenile herring, and indirectly by adult herring influencing the biomass of euphausiids (Table 4.2). Both herring TD impacts are equally strong in influencing copepod annual production or mean biomass. In contrast to model results, Koslow (1983) noted that reductions in herring and mackerel biomass, in the North Atlantic during the late 1960s, did not result in increased copepod biomass. Koslow's (1983) size-based feeding model predicted that copepod biomass should have increased 170 concurrently with the decline in planktivore biomass. Koslow (1983) argues that prey limitation or the effects of physical processes rather than alternate vertebrate or invertebrate predators is likely responsible for the unexpected response. This conclusion is supported by comparing the results from the 1972-90 simulations and upwelling rate experiments (Figure 4.2). Other studies have indicated that small fishes can have noticeable TD impacts on marine copepods, but these have been restricted to systems with relatively simple physical properties, like shallow bays (e.g., Kimmerer and McKinnon 1989). Euphausiids have a strong TD impact on annual copepod production (TD #6) . The strength of this TD impact is revealed in the correlation between euphausiid and copepod production (Figure 4.6). Pillar et al. (1990) similarly indicate that euphausiids play an important regulatory role in the production (and biomass) of mesozooplankton in the southern Benguela. The authors indicate that this trophic interaction is stronger than the euphausiid-phytoplankton pathway. In the North Sea, Roff et al. (1988) determined that euphausiid predation influenced copepod biomass, but they could not detect any TD effect on copepod production. It is also known that marine copepod populations can be significantly effected by predation from other invertebrates. In the summer in the Strait of Georgia, for example, ctenophores have large predatory impacts on copepods (e.g., Greve and Parsons 1977). In the Eddy region, ctenophores occasionally occur in large numbers coincident with intrusions of warm oceanic waters onto the shelf. The TD impacts of ctenophores may be restricted to smaller copepod stages or genera as suggested by Greve and Parsons (1977), because larger copepods tear the tentacles of the predatory ctenophores. The relative TD impacts of copepods and euphausiids on diatoms are both considered moderate (TD # 2 and TD #4 , respectively). Although the model euphausiids have 2-5 times lower ingestion and about 5 times lower growth, than copepods, their TD impacts on diatoms are equivalent. In part, this equivalence is related to the strong TD Copepod production (g C/m /y) Figure 4.6. Annual copepod production plotted against annual euphausiid production (r = -.71). Data from the 1972-90 simulation. 172 impact of euphausiids on the copepods (TD #6; Figure 4.1). Simulations using real world oceanic variability (Figure 4.4) however, indicate that even if TD interaction #6 is relatively weak, TD #2 is still limited in strength because of the temporal numerical lag between faster growing diatoms and the substantially slower growing mesozooplankton (Chapter 2, Kiorboe 1993). Ohman (1986) indicates that euphausiids are only one of several possible marine invertebrates that have significant TD impacts on coastal copepods (e.g., chaetognaths, ctenophores, and other gelatinous zooplankton). From the above discussion, it becomes apparent that intermediate trophic level marine plankton are influenced simultaneously by interactions between resources and predators. For instance, model copepod production is determined by strong BU effects of diatoms, in combination with strong TD impacts from euphausiids and herring, and by physical processes, like offshore transport. The interplay between relatively high copepod growth rate and several strong "forcing" functions leads to high variability in copepod biomass, as determined in the sensitivity analyses (See Chapter 2). This may explain the frequent inability to couple juvenile fish survival/growth to copepod production dynamics in the real world (Cushing 1990). Euphausiids are also forced by both strong BU and TD interactions, but they exhibit relatively low production variability. This is because maximum potential euphausiid production is determined by BU effects, while realized production is strongly controlled by TD effects of predators like hake (see Carpenter et al. 1987). The reason why euphausiids are more susceptible to predators than copepods is related to the fact that euphausiids have lower growth capabilities (e.g., lower ingestion and growth rates, longer turn-over times), and thus euphausiids are unlikely to "outgrow" their predators. Predator behaviour like size-selection would also allow for rapid adaptations to variable and aggregated euphausiid biomass. These properties should enhance the predictability in estimating production for fishes primarily dependent upon euphausiids. Pacific herring production is also strongly effected by BU/TD interplay. 173 Conditions enhancing herring growth, like high euphausiid production, are combined with increased predation pressure from hake. Herring are only moderately effected by increased euphausiid production because more hake remained in the Eddy region (Figure 2.22; Figure 4.4). Thus for herring, the benefits of increased euphausiid production are potentially offset by costs of increased hake predation (Figure 3.13). The seasonal movements and growth of herring in the Eddy region during the upwelling season support these simulation results (Ware and McFarlane 1994). 4.3.3 Temporal variability in BU/TD interactions Temporal variability in the BU/TD interactions discussed above are not frequently studied because extensive data sets are required. The simulation model however, provides an opportunity to evaluate the relative importance of BU versus TD processes in the coastal marine system over the year. In this section, I present the results from model experiments to highlight some of the possible seasonal variability in BU/TD processes. In the model experiments, variable upwelling rate produces 2-5 (4-10) times greater variability in diatom biomass (production), during the upwelling season than do TD effects originating from variable hake biomass (Figure 4.7). Only in September, are TD effects by zooplankton equivalent to BU effects in generating diatom variability. TD impacts of model copepods on diatoms are important in September because diatom growth slows due to nitrogen limitation (reduced upwelling), and because relatively high grazing copepod grazing biomass carries over from summer. The latter may be considered a form of predatory inertia (sensu Carpenter et al. 1985). High phytoplankton production and biomass also occurs throughout the upwelling season on the Washington Shelf, ultimately reducing the lag response of grazers, and thus maintaining large zooplankton stocks capable of consuming high average levels of phytoplankton (Landry and Lorenzen 1989). If we compare seasonal variations in BU versus TD effects on copepods using model experimental results, we would be mislead because of the absence of physical 60 ^ 5 0 >9 0 s > O 40 biomass o E 20 0 * • > (0 S,o 0J A "•"Upwelling rate •*• Hake biomass -k 4 A i -/ \ / / \ \ _ - - - - • — — * \ \ ~~ / / ^ \ ~ K i i i i i i ' ** ^ ^ ^ ^ ^c ^s v^ *? o * ^ <?< Month 16 14 12 O 2 10 > O 8 Diatom 01 4 2 n B •*• biomass "*" production --i i i -i /_ / ^ v / -—^"^ \ \ 1 1 1 1 1 I 1 1 ** <? ^ ^ *? .»* ^ o° *° 16 14 12 10 8 6 4 Month Figure 4.7. Panel A: The coefficient of variation (CV) in monthly diatom biomass generated in the upwelling rate and hake model experiments. Panel B: The ratio of CVs from experiments for both diatom biomass and production. 175 processes acting on the copepods (compare Figures 4.2 and 4.5). A more reasonable evaluation of seasonal variability in TD impacts on copepods is to remove the BU effects by regression, and then plot the residuals against predator biomass (e.g., euphausiids; after Carpenter et al. 1985). If TD influences are important, then large positive residuals will be associated with low predator biomass, while large negative residuals will be associated with high predator biomass (i.e., a significant negative correlation will occur). This approach assumes that the effects of food-web interactions are independent of those due to BU processes (Carpenter et al. 1985). The results from the above analysis indicate that TD impacts of euphausiids on copepods are significant in summer (June to August), while TD impacts from juvenile herring are important in autumn (Table 4.4). BU processes seem to be most important to copepod production in spring. Roff et al. (1988) also indicate that North Sea copepods experience "seasonally sequential" prey and predator limitation. Copepods are prey limited in the spring and early summer, before becoming predator limited from July onwards; euphausiids had the largest TD impact on copepods in late autumn/early winter. Copepods in the Baltic Sea also experience seasonally variable TD impacts from zooplanktivores, with the late summer and autumn being most intense (Rudstram et al. 1992). Model euphausiids also experience seasonally sequential prey and predator limitation. Specifically, BU effects generate only slightly greater biomass variability in June, September, and October (1-1.6 times), while TD effects generate between 1.3-2.5 times greater biomass variability in July, August, November, and December (Figure 4.8). An evaluation of the bivariate plots of residuals from the 1972-90 seasonal prey-resource regressions versus predator biomasses confirms that TD effects on euphausiids are most important in summer (Table 4.4). Herring production does not show any seasonal variability in the relative importance of BU versus TD effects. Herring production consistently shows between 3 Table 4.4. Correlations between prey-resource regression residuals and predators, for plankton biomass and production estimated in each year during the 1972-90 simulation For N = 19, the correlation coefficient has to exceed ± 0.575 (**) and ± 0.456 (*) to be significant at the 1% and 5% levels, respectively. BIOMASS Winter Spring Summer Autumn Annual DIATOMS Cop +.04 +.20 +.64** +.01 +.41 Euph +.43 +.32 -.11 +.73** +.27 COPEPODS Euph +.73 ** -.01 -.45* -.27 -.57** Juv. herr -.43 -.43 +.34 +.01 +.06 EUPHAUSIIDS Adl. herr +.16 -.32 +.09 -.30 -.16 Hake +.15 NA -.46* +.43 +.02 PRODUCTION Winter Spring Summer Autumn Annual DIATOMS Cop +.56* +.14 +.62** +.06 +.37 Euph +.71** +.50* -.13 +.69** -.14 COPEPODS Euph +.60** +.12 -.57** -.24 -.67** Juv. herr -.30 -.41 +.31 -.29 +.23 EUPHAUSIIDS Adl. herr +.06 +.38 -.08 -.26 -.28 Hake +.38 NA +.02 +.68** +.33 177 e.\i ^ ?: 15 o </) il> (0 £ .2 10 n •a w 3 « 5 J= ° a 3 UJ 0J "•"Upwelling rate "*" Hake biomass / ^~~K / ^ \ t A A i £ 1 1 1 1 1 1 Month 20 15 10 * * « * ^ ^ ^ >** ^ ^ #* Of5" «5>* <>* 2.5 2.0 O 13 Q i _ > 1 . 5 2 "3 3 1 0 £ a 3 UJ 0.5 n n B "*• Biomass "*-Production i i i \ — \ / A\ \ /* \ \ i i i i i i i 2.5 2.0 - 1.5 1.0 0.5 * ^ ^ J ^ op o* *°A <r 0.0 Month Figure 4.8. Panel A: The coefficient of variation (CV) in monthly euphausiid biomass generated in the hake biomass and upwelling rate experiments. Panel B: The ratio of CVs from model experiments for euphausiid biomass and production. 178 and 10 times greater variability when forced by TD effects than BU effects, from July to the end of the year (Figure 4.9). These simulation results are partly due to limitations concerning herring-prey feeding pathways included in the model, and to the dynamic feeding interactions between hake, herring, and euphausiids. The above discussion concerning the relative seasonal importance of BU versus TD processes on plankton and fish production assumes that the "food-web" remains constant throughout the year. This assumption is dubious because of known changes in phytoplankton and zooplankton species composition, associated with changes from upwelling to downwelling regimes or vice versa (Mackas 1992). Accompanying these species changes are differences in relative cell and body size, and thus the associated trophodynamical phasing properties of the plankton community (sensu Parsons 1988). Oceanic waters of the central Pacific, for example, are typically dominated by small phytoplankton (< 10 |im), while zooplankton biomass is generally dominated by smaller zooplankton (e.g., microzooplankton; Frost 1987; Parsons and Lalli 1988). Since the zooplanktonic predators have turnover times that are similar to the primary producers, there is close trophodynamical coupling, and little if any seasonal increase in phytoplankton biomass (Evans and Parslow 1985; Frost 1987; Parsons and Lalli 1988). In shelf waters during the downwelling period then, the neritic plankton community is replaced primarily with an oceanic community that is more likely to be TD controlled. An analogous situation occurs with changes in nutrient status in lakes. The interplay between BU/TD interactions in plankton and fish communities changes, such that in oligotrophic lakes TD effects are not as strongly buffered, and thus zooplankton to phytoplankton interactions are significant (McQueen et al. 1986). Shifts in plankton community structure and associated variability in the relative importance of BU/TD interactions, depending upon upwelling and downwelling conditions, highlights an additional consideration. There may be shifts in plankton community structure during the upwelling season. Davis (1987), for example, notes that 179 35 30 > 25 o » «) 20 TO £ I" o> •E io 01 z "•" Upwelling rate •*• Hake biomass Ok A A * * ^ « * * * ^ ^ ^ ^ ^ of O* ^ jfi Month B 1.0 0.8 O 33 « > o O) •§0.4 0) X 0.2 0.6 0.0 Biomass Production ^ ^ • Month Figure 4.9. Panel A: The coefficient of varation (CV) in monthly herring biomass generated in the hake biomass and upwelling rate experiments. Panel B: The ratio of CVs from model experiments for herring biomass and production. 180 there is a strong seasonal component in the relative dominance of large (e.g., Calanus) versus small copepods (e.g.,Acartia) in temperate shelf systems. Large biomasses of large copepods are associated with spring diatom blooms, while smaller copepods increase in the warmer stratified summer (See also Parsons and Lalli 1988). Results from a simulation model describing plankton dynamics in the Benguela upwelling system also indicate that the plankton community structure, and its trophodynamic properties, change in response to upwelling and non-upwelling conditions (Moloney 1992). It may be that the relative frequency of upwelling during the summer determines the plankton community structure and the intensity of planktonic B U / T D interactions. In addition to shifts in plankton community structure, interannual changes in the interactions between BU/TD processes higher in the food-web are possible. At least two contrasting nektonic summertime food-webs have been found to be dependent upon coastal oceanographic conditions (Brodeur and Pearcy 1992). During the strong upwelling years, Brodeur and Pearcy (1992) found that trophic diversity of prey is low, with euphausiids being the dominant organism consumed. The model indicates that TD impacts by fish on zooplankton are strong under these conditions. In contrast, in 1983 a year of reduced upwelling, there is a reduction in euphausiids in fish diets and an increase in prey diversity. It is expected that TD impacts of fishes in this instance are lessened because they would be dispersed in response to reduced biomasses and aggregations of euphausiids. In warm years feeding interactions become uncoupled further up the food-web. To sum up, this Chapter has evaluated the relative importance of resource limitation versus predation in influencing trophic level production of a coastal food-web. It is clear from the model experiments that oceanic variability (e.g., nutrient input) is the critical factor influencing the BU/TD interactions. Oceanic variability determines the strength of the BU/TD interactions by effecting extant trophodynamic phasing, and/or by effecting the possible plankton/fish community structure. 181 GENERAL DISCUSSION The trophodynamics model In the first Chapter of my dissertation I present the structure of a simulation model that describes the feeding interactions among dominant pelagic organisms in a coastal temperate upwelling system. I recognize that the simulated food-web is a simplification of the real world. However, it is not unreasonable given that: 1) diatoms are dependent upon certain coastal upwelling conditions for rapid growth (e.g., high nitrogen concentrations, turbulence), 2) zooplankton are size-selective grazers, and 3) pelagic marine fishes feed on aggregations of meso/macrozooplankton. As several biological oceanographers have recently noted, these characteristics are what make coastal regions high in fish production (Parsons 1979; Cushing 1989; Legendre 1990; Mann 1992; Kiorboe 1993). There is one major change in the model food-web structure that I would make to more fully evaluate the response of the production system to oceanic variability. The phytoplankton could be divided into nano versus net, and copepods into small (e.g., Acartia) versus large (e.g., Calanus) genera. These divisions would help elucidate system production dynamics because it is recognized that in temperate coastal regions larger neritic copepods are associated with diatom blooms, while smaller genera occur with nanophytoplankton. In turn, these pairings are primarily dependent upon temporal variability in nutrient input (Davis 1987; Legendre 1990; Moloney 1992). There are however, two major factors presently limiting the division of model plankton. There is a definite lack of data concerning basic biology (e.g., feeding habits) and seasonal abundances of smaller plankton in the La Perouse region. There is also little understanding of how changes in local oceanographic processes can effect the dynamics and interactions of the different plankton communities. The latter point is not peculiar to the La Perouse region as emphasized by the direction of the international program on Global Ocean Ecosystem Dynamics (GLOBEC) in developing a generic turbulence-closure 182 model of mixed layer depth "as a vehicle for modelling the seasonal cycles of phytoplankton/zooplankton coupling". An enhanced understanding of temporal variability in physical oceanographic processes in the Eddy region like, mixed layer dynamics, sources of nutrient input (e.g., Mackas et al. 1987), and interactions with surrounding oceanic subregions, is critical information needed before increasing the trophodynamical complexity of the model plankton. Evaluation of model assumptions and structure An overview of the standard run of the Juan de Fuca Eddy trophodynamics model is presented in Chapter 2. I initially describe the calibration and corroboration of the trophodynamics model to the available empirical data. The good agreement between the standard run and the empirical data is a measure of the validity of the simplifying assumptions used. However, it is clear from calibrating the model to available data that several key biological processes need to be better understood and quantified. The most important process requiring immediate attention is the apparent imbalance between fish consumption and euphausiid production. The simulation results highlight the requirement for improved seasonal estimates of zooplankton biomass, particularly in late spring and early summer, evidence of euphausiid import and aggregation mechanisms, and basic life history information for Thysanoessa spinifera. With respect to the latter point, I am surprised by how little information is available for T. spinifera given the relative importance of this animal to most adult coastal marine fishes (e.g., Tanasichuk et al. 1991; Brodeur and Pearcy 1992). It is noteworthy that Ron Tanasichuk (DFO, Nanaimo) is presently conducting a multiple year study of T. spinifera population dynamics in Barkley Sound, adjacent to La Perouse Bank. Another important process identified in the model that needs to be more fully addressed is the dynamical interaction between euphausiids and hake. Information is needed to better understand how associations of euphausiids and hake on the southern BC 183 shelf are influenced by coastal oceanographic properties and dynamics. An understanding of how local and regional hake migration characteristics respond to oceanographic and biological factors is also required. Data collection concerning migration characteristics, like arrival timing, may prove elusive because of the inability to tag hake. The final section of Chapter 2 investigates the relative sensitivity of the trophodynamics model to parameter variability, and assumptions concerning hake migration. The results illustrate that the observed ranges of starting concentrations of state variables have relatively small influences on annual plankton or fish production or mean annual biomass. In addition, output obtained after one annual simulation is an accurate realization of model parameters, processes, and functions, when compared to longer simulations. The sensitivity analyses also indicate that the arrival timing and biomass of hake may be important features influencing system production dynamics. Temporal variability in production Several recent studies (e.g., Brodeur and Ware 1992, Beamish and McFarlane 1993) have suggested that ocean-climate interactions in the NE Pacific Ocean indirectly and primarily determine zooplankton biomass or production, and ultimately pelagic fish production. This study represents one of the few quantitative attempts at evaluating these interactions in coastal regions. I have used the simulation model, and observed interannual variability in environmental factors and fish biomass, to hindcast plankton production and fish production from 1972 to 1990. The main objective of the simulations is to provide insights into potential mechanisms or processes influencing short and long-term production variability. From an evaluation of the effects of simplifying assumptions or parameter values on simulated plankton and fish production it is apparent that while absolute values of annual production may differ, the interannual and longer-term patterns remain similar. This is because the majority of the parameters have little or no impact on the relative 184 trophodynamical phasing properties of the plankton (sensu Parsons 1988). Changing the copepod offshore transport rate or the euphausiid import, for example, has large effects on annual zooplankton production, but diatoms are relatively unaffected. In contrast, reducing nutrient input or sunshine, effectively slows diatom growth, and results in substantial changes in the interannual production patterns. These results highlight the relative importance of physical oceanographic variability in overriding biotic factors and determining system production. The model results also indicate that interannual and longer-term variability in coastal plankton production occurs, and that some years may favour copepod production versus euphausiid production and vice versa. The trends in annual plankton production are determined primarily by the dynamics in spring and summer coastal upwelling, which in turn are effected by atmospheric pressure systems. Simulated hake production is mainly influenced by summer euphausiid production, while measured herring condition is related to spring and autumn zooplankton production. An evaluation of the temporal variability in transfer efficiency indicates that seasonal and interannual aspects need to be considered when using plankton production to estimate fish production in dynamic coastal upwelling systems. Bottom-up versus top-down processes Chapter 4 of my thesis evaluates the relative importance of resource limitation (bottom-up) versus predation (top-down) processes in a coastal food-web. It is apparent from the simulation results and the few empirical studies conducted in marine environments, that oceanic variability (e.g., nutrient input, Ekman transport) plays the dominant role in influencing BU/TD interactions. The relative interplay among BU/TD processes changes because oceanic variability influences trophodynamical phasing (Parsons 1988). Changes in upwelling rate, for example, influences diatom production dynamics which are subsequently transferred up the "food-web". In contrast, the strength of the top-down cascade from hake to diatoms is biologically determined at the 185 euphausiid-copepod trophic link. However, even if the copepod-euphausiid feeding link is strong, top-down effects are effectively lost by real world coastal oceanic (BU) variability influencing the copepod-diatom interaction (Denman et al. 1989). Unlike eutrophic lakes where variation around BU processes can typically be explained by TD forces (McQueen et al. 1986), oceanic factors determine the strength of interplay of the myriad of B U / T D processes in productive coastal marine systems (Bailey and Francis 1986; Denman et al. 1989). Oceanic variability may also determine the strength of BU/TD interactions by changing plankton community structure, and thus the transfer of productivity between lower trophic levels, and the relative success in predicting higher trophic level productivity. Overview Overall my dissertation research represents the only published simulation model describing temporal variability in plankton and fish feeding/production dynamics in an upwelling region of the Coastal Upwelling Domain. In addition, the estimates of plankton and fish production are the first published for the continental shelf off southern Vancouver Island, B.C.. The approach differs from several published models developed for other upwelling systems (e.g., Moloney 1992) in that a multi-species model, from plankton to fish, is forced by oceanic variability and fish predation. Thus, most of the model processes and functions are linked to the state of the oceanic system rather than being dependent on time. This approach produces more realistic simulations and a better understanding of system functioning. Several additional novel aspects of the modelling approach are worth noting. The model includes the effects of both local and regional oceanic forcing on plankton and fish trophodynamics, and thus provides a holistic view of the systems production dynamics. Most studies concentrate only on events or processes occurring locally. It is apparent however, that large-scale oceanic processes (e.g., changes in atmospheric systems) play an extremely important role in shaping the system dynamics. 186 Another novel approach used in this study is the inclusion of important interactions between fish and their zooplanktonic prey. My study indicates that empirically derived functions, where fish predator migrations are linked to zooplanktonic prey biomass, should be included in simulation models to better understand the production dynamics of coastal upwelling systems containing migratory planktivores. The modelling approach discussed above has provided two major insights into the functioning of coastal marine production systems. By using the model to hindcast plankton and fish production, I have clearly demonstrated both long and short term variability in annual production. The short-term variability, particularly in the copepods, highlights why it is so difficult to quantify long lasting zooplankton-fish relationships. This result of interannual variability in the copepods supports GLOBEC's recent initiative of understanding the effects of physical process on predator-prey interactions and population dynamics of zooplankton and their relation to ocean ecosystems in the context of the global climate system and anthropogenic change. An equally important simulation result is that real-world oceanic variability is the dominant factor in determining production dynamics and potential of productive coastal systems. By contrast, recent studies of bottom-up versus top-down interactions in eutrophic lakes indicate that phytoplankton are influenced by both resources and predators, while upper trophic levels are primarily influenced by predators (McQueen et al. 1989). The evaluation of resource versus predator limitations using the simulation model has clearly added to our understanding of the myriad of processes influencing productive coastal systems. Future directions My impression from reading the ecological modelling literature is that simulation models are frequently developed for specific problems, and then disregarded after some short time. It is apparent however, that if we are to systematically evaluate the relative importance of oceanic factors on coastal plankton and fish production, then we must 187 continue to develop and modify existing simulation models. The Juan de Fuca Eddy trophodynamics model could be expanded to take into account several of the processes discussed above, and used to investigate the longer-term (decadal) variability in system production. For instance, the model could be used to compare the carbon flow of the current pelagic system off Vancouver Island with a reconstruction of the historical system. The La Perouse upwelling system consisted of herring, hake, and dogfish in addition to a relatively large biomass of Pacific sardine {Sardinops sagax), in the 1930s and 1940s. The abundant migratory Pacific sardine resulted in a summer/fall pelagic fishery in British Columbia averaging about 40 kt y"1 (Murphy 1966). Large biomasses of sardines in the La Perouse region are thought to have negatively impacted the growth and recruitment success of the Pacific herring because of increased competition for zooplanktonic prey (Ware and Thompson 1989). The sudden disappearance of sardines from the La Perouse region in the mid 1940s due to the collapse of the California stock, may have had substantial negative impacts (reduced efficiency) and positive impacts (less competition for herring) on carbon flow through the whole system (Ware 1991). A thorough evaluation of these hypothesized species interactions is important given that the California sardine population is currently expanding at a rate of about 30% per year (Smith et al. 1992). In fact during the summers of 1992 and 1993, adult sardines that had migrated from California were caught off southern Vancouver Island by DFO research vessels for the first time in over 45 years (D. Ware, Pacific Biological Station, Nanaimo, B.C.).. In summary, I have discussed a realistic simulation model capable of providing a holistic view of how oceanic variability and feeding interactions influence pelagic plankton and fish production. The technique of simulation modelling has proven to be an effective method for synthesizing a large amount of data and information, and for identifying knowledge gaps. More importantly, the trophodynamics model has shown to be an effective quantitative tool for generating annual estimates of production, and for testing ecological hypotheses about the structure and functioning of upwelling ecosystems. 188 LITERATURE CITED Alton, M.S. and C.J. Blackburn. 1972. Diel changes in the vertical distribution of the euphausiids, Thysanoessa spinifera Holmes and Euphausia pacifica Hansen in coastal water of Washington. Calif. Fish and Game. 58:179-190. Anderson, V. and P. Nival. 1991. A model of the diel vertical migration of zooplankton based on euphausiids. J. mar Res. 49:153-175. Armstrong, M.J., A.G. James, and E.S. Valdes-Szeinfeld. 1991. Estimates of annual consumption of food by anchovy and other pelagic fish species off South Africa during the period 1984-1988. S. Afr. J. mar. Sci. 11:251-266. Bailey, K.M.and R.C.Francis . 1986. Environment and population dynamics of Pacific whiting, Merluccius productus. INPFC Bull. 45:317-333. Bailey, K.M., R.C. Francis, and P.R. Stevens. 1982. The life history and fishery of Pacific whiting, Merluccius productus. CalCOFI Rep. 23:81-98. Bakun, A. 1973. Coastal upwelling indices, west coast of North America 1946-71. NOAA technical report NMFS SSRF-671. 13 p. Bakun, A. 1990. Global climate change and intensification of coastal ocean upwelling. Science 247:198-201. Banse, K. and S. Mosher. 1980. Adult body mass and annual production/biomass relationships of field populations. Ecological Monographs 50:355-379. Barange, M. and S.C. Pillar. 1992. Cross-shelf circulation, zonation and maintenance mechanisms of Nytiphanes capensis and Euphausia hanseni (Euphansiacea) in the northern Benguela upwelling system. Cont. Shelf Res. 9:1027-1042. Baumgartner, T.R., A. Soutar, and V.F. Bartrina. 1992. Reconstruction of the history of Pacific sardine and Northern anchovy populations over the past two millennia from sediments of the Santa Barbara Basin, California. CalCOFI Rep 33:24-40. Beamish, R.J. . 1994. Climate change and exceptional fish production off the west coast of North America. Can. J. Fish. Aquat. Sci. 51: In Press. Beamish, R.J. and G.A. McFarlane. 1985. Pacific whiting, Merluccius productus, stocks off the west coast of Vancouver Island, Canada. Mar. Fish. Rev. 47:75-81. Beamish, R.J. and D. Bouillon. 1993. Pacific salmon production trends in relation to climate. Can. J. Fish. Aquat. Sci. 50:1002-1016. Bernal, P.A. 1981. A review of the low-frequency response of the pelagic ecosystem in the California Current. CalCOFI Rep. 22:49-64. Bollens, S.M., B.W. Frost, and T.S. Lin. 1992. Recruitment, growth, and diel vertical migration of Euphausia pacifica in a temperate fjord. Mar. Biol. 114:219-228. Brinton, E. 1967. Vertical migration and avoidance capability of euphausiids in the California Current. Limnol. Oceanogr. 12:451-483. 189 Brinton, E. and A.W. Townsend. 1981. A comparison of euphausiid abundances from bongo and 1-mCalCOFI nets. CalCOFI Rep. 22:111-125. Brodeur, R.D. and W.G. Pearcy. 1992. Effects of environmental variability on trophic interactions and food web structure in a pelagic upwelling ecosystem. Mar. Ecol. Prog. Ser. 84:101-119. Brodeur, R.D. and Ware, D.M. 1992. Long-term variability in zooplankton biomass in the subarctic Pacific Ocean. Fish. Oceanogr. 1:32-38. Brown, P.C., S. J. Painting, and K.L. Cochrane. 1991. Estimates of phytoplankton and bacterial biomass and production in the northern and southern Benguela Ecosystems. S. Afr. J. mar. Sci. 11:537-564. Carpenter, S.R. J.F. Kitchell, and J.R. Hodgson. 1985. Cascading trophic interactions and lake productivity. Bioscience 35:634-639. Carpenter, S.R., J.F. Kitchell, J .R.Hodgson, P.A.Cochrane, J.J. Elser, M.M.Elser, D.M. Hodge, D. Kretchmer, X. He, C.N. von Ende. 1987. Regulation of lake primary productivity by food web structure. Ecology 68:1863-1876. Chavez, F.P., R.T. Barber, P.M. Kosro, A. Huyer, S.R. Ramp, T.P. Stanton, and B. Rojas de Mendiola. 1991. Horizontal transport and the distribution of nutrients in the coastal transition zone off northern California: Effects on primary production, phytoplankton biomass and species composition. J. Geoph. Res. 96:14833-14848. Chelton, D.B., P.A. Bernal, and J.A. McGowan. 1982. Large-scale interannual physical and biological interaction in the California Current. J. mar. Res. 40:1095-1125. Chisholm, S.W. and F.M.M. Morel. 1991. What controls phytoplankton production in the nutrient-rich areas of the open sea? Limnol. Oceanogr. 36:1507-1970. Cochrane, N.A., D. Sameoto, A.W.Herman, and J. Neilson. 1991. Multiple-frequency acoustic backscattering and zooplankton aggregations in the inner Scotian Shelf basins. Can. J. Fish. Aquat. Sci. 48:340-355. Colebrook, J.M. 1977. Annual fluctuations in biomass of taxonomic groups of zooplankton in the California Current, 1955-59. Fish. Bull. 75:357-368. Colebrook, J.M. 1984. Continuous plankton records: relationships between species of phytoplankton and zooplankton in the seasonal cycle. Mar. Biol. 84:261-265. Conover, R.J. 1978. Transformation of organic matter, p. 221-499. In Kinne, O. (ed) Marine Ecology Vol 4. Dynamics, Wiley, New York. Cooney, R.T. and K.O. Coyle. 1982. Trophic implications of cross-shelf copepod distributions in the Southeastern Bering Sea. Mar. Biol. 70:187-196. Corkett, C.J. and LA. McLaren. 1978. The biology of Pseudocalanus. Adv. Mar. Biol. 15: 2-211. Crawford, W.R. and R.K. Dewey. 1989. Turbulence and mixing. Sources of nutrients on the Vancouver Island continental shelf. Atmosphere-Ocean, 27:428-442. 190 Cury, P. and C. Roy. 1989. Optimal environmental window and pelagic fish recruitment success in upwelling areas. Can. J. Fish. Aquat. Sci. 46:670-680. Cushing, D.H. 1971. Upwelling and the production of fish. Adv. mar. Biol. 9:255-334. Cushing, D.H. 1975. Marine Ecology and Fisheries. Cambridge University Press, Cambridge. 278 p. Cushing, D.H. 1978. The upper trophic levels in upwelling areas, p. 101-110. In B. Boje and M. Tomczak (eds), Upwelling Systems. Springer-Verlag, Berlin. Cushing, D.H. 1989. A difference in structure between ecosystems in strongly stratified waters and in those that are only weakly stratified. J. Plankton Res. 11:1-13. Cushing, D.H. 1990. Plankton production and year-class strength in fish populations: An update of the Match/Mismatch Hypothesis. Adv. Mar. Biol. 26:249-293. Daly, K.L. and M.C. MaCaulay. 1988. Abundance and distribution of krill in the ice edge zone of the Weddell Sea, austral spring 1983. Deep-Sea Res. 35:21-41. Davis, C.S. 1987. Components of the zooplankton production cycle in the temperate ocean. J. mar. Res. 45:947-983. Day, D.S. 1971. Macrozooplankton and small nekton in the coastal waters off Vancouver Island (Canada) and Washington, Spring and fall of 1963. NOAA Special Scientific Report, Fisheries No. 619. 16 p. DeAngelis, D.L., P.J. Mulholland, A.V. Palumbo, A.D. Steinman, M.A. Huston, and J.W. Elwood. 1989. Nutrient dynamics and food-web stability. Ann. Rev. Ecol. Syst. 20:71-95. Denman, K.L., D.L. Mackas, H.J. Freeland, M.J. Austin, and S.H. Hill. 1981. Persistent upwelling and mesoscale zones of high productivity off the west coast of Vancouver Island, Canada, p. 514-521. In Richards, F.A. (ed) Coastal Upwelling, Coastal and Estuarine Sciences, American Geophysical Union. Washington. Denman, K.L., H.J. Freeland, and D.L. Mackas. 1989. Comparisons of time scales for biomass transfer up the marine food web and coastal transport processes. Can. Spec. Publ. Fish. Aquat. Sci. 108:255-264. Dickson, R.R, P.M. Kelly, J.M. Colebrook, W.S. Wooster, and D.H. Cushing. 1988. North winds and production in the eastern North Pacific. J. Plankton Res. 10:151-169. DiStefano III, J.J., A.R. Stubberud, and I.J. Williams 1967. Theory and problems of feedback and control systems, p. 179-186. Schaum Publishing Co., New York. Dunbrack, R. and D.M.Ware. 1986. Particle size spectrum estimation of the standing stock of pelagic fish. Can. Tech. Rept. Fish. Aquat. Sci 1470:1-3. Eppley, R.W. 1972. Temperature and phytoplankton growth in the sea. Fish. Bull. 70:1063-1085. Evans, G.T. and J.S. Parslow. 1985. A model of annual plankton cycles. Biol. Oceanogr. 3:327-347. 191 Evans, G.T. and P. Pepin. 1989. Potential for predicting plankton populations (and fish recruitment) from environmental data. Can. J. Fish. Aquat. Sci. 46:898-903. Everson, I. and D.G. Bone. 1986. Effectiveness of the RMT8 system for sampling krill (Euphausia superba) swarms. Polar Biology 6:83-90. Forbes, J.R., S.L. Buckingham, and A.T. Earmme. 1987. Phytoplankton productivity experiments in British Columbia coastal waters, 1986. Can. Data Rep. Hydrogr. Ocean Sci. 56: 169 p. Francis, R.C. 1983. Population and trophic dynamics of Pacific hake (Merluccius productus). Can. J. Fish. Aquat. Sci. 40:1925-1943. Franz, H.G., J.P. Mommaerts, and G. Radach. 1991. Ecological modelling of the North Sea. Netherlands Journal of Sea Res. 28:67-140. Freeland, H.J. and K.L. Denman. 1982. A topographically controlled upwelling centre off southern Vancouver Island. J. Marine Res. 40:1069-1093. Freeland, H.J. 1991. Sea surface temperatures along the coast of British Columbia: Regional evidence for a warming trend. Can. J. Fish. Aquat. Sci. 47:346-350. Frost, B.W. 1987. Grazing control of phytoplankton stock in the open subarctic Pacific Ocean: a model assessing the role of mesozooplankton, particularly the large calanoid copepods Neocalanus spp. Mar. Ecol. Prog. Ser. 39:49-68. Fulton, J., M.N. Arai, and J.C. Mason. 1982. Euphausiids, coelenterates, ctenophores, and other zooplankton from the Canadian Pacific coast ichthyoplankton survey, 1980. Can. Tech. Rep. Fish. Aquat. Sci. No. 1125: 75 p. Garrison, D.L. 1976. Contribution of the net plankton and nannoplankton to the standing stocks and primary productivity in Monterey Bay, California during the upwelling season. Fish. Bull. 74:183-193. GLOBEC Report No. 2. 1993. Population dynamics and physical variability. Published by the Scientific Committee on Oceanic Research, The Johns Hopkins University, Baltimore. 104 p. Greve, W. and T.R. Parsons. 1977. Photosynthesis and fish production: Hypothetical effects of climatic change and pollution. Hlgolander wiss. Meeresunters 30:666-672. Harris, G.P. and F.B. Griffiths. 1987. On means and variances in aquatic food chains and recruitment to the fisheries. Freshwater Biology 17:381-386. Harris, G.P., F.B. Griffiths, and L.A. Clementson. 1992. Climate and fisheries off Tasmania-interactions of physics, food chains and fish. S. Afr. J. mar Sci. 12:107-121. Harrison, P.J. and D.H. Turpin. 1982. The manipulation of physical, chemical and biological factors to select species from natural phytoplankton communities, p. 275-289. In Grice, G. and M.R. Reeve (eds) Marine Mesocosms. Biological and Chemical Research in Experimental Ecosystems. Springer- Verlag, New York. Harrison, P.J., P.A. Thomson, and G.S. Calderwood. 1990. Effects of nutrients and light 192 limitation on biochemical composition of phytoplankton. J. Appl. Phycol. 2:45-56. Heath, W.A. 1977. The ecology and harvesting of euphausiids in the Strait of Georgia. Ph.D. Thesis University of British Columbia. 187 p. Hickey, B.M. 1979. The California Current System-hypotheses and facts. Prog. Oceanogr., 8:191-279. Hill, S., K. Denman, D. Mackas, and H. Sefton. 1982. Ocean ecology data report: coastal water off southwest Vancouver Island. Spring and summer 1979. Can. Data Rep. Hydrogr. Ocean Sci. 3:118 p. Hofmann, E.E. and J.W. Ambler. 1988. Plankton dynamics on the outer southeastern US continental shelf. Part II: A time-dependent biological model. J. mar. Biol. 46:883-917. Hollowed, A.B. 1990. Recruitment of marine fishes in the northeast Pacific Ocean in relation to interannual variations in the environment. Ph. D. Dissertation. University of Washington, Seattle, WA. 281 p. Hollowed, B.A. and W.S. Wooster. 1991. Year class variability of marine fishes and winter ocean conditions in the northeast Pacific Ocean. ICES 1991 Symposium paper No. 3. 26 p. Houde, E.D. and S.A. Berkley. 1981. Food and growth of juvenile herring Clupea harengus pallasi, in CEPEX enclosures, p. 239-248. In G.D. Grice and M.R. Reeve (eds), Marine Mesocosms - Biological and chemical research in experimental ecosystems. Springer-Verlag. Berlin. Hourston, A.S. and C.W. Haegele. 1980. Herring on Canada's Pacific Coast. Can. Spec. Publ. Fish. Aquat. Sci. 48:23 p. Hovekamp, S. 1989. Avoidance of nets by Euphausia pacifica in Dabob Bay. J. Plankton Res. 11:907-924. Huntley,M. and C. Boyd. 1984. Food-limited growth of marine zooplankton. Am. Nat. 124:455-478. Hutchings, L. S.C. Pillar, and H.M. Verheye. 1991. Estimates of standing stock, production and consumption of meso- and macrozooplankton in the Benguela ecosystem. S. Afr. J. mar. Sci. 11:499-512. Huyer, A. 1983. Coastal upwelling in the California Current System. Progr. Ocean. 12:259-284. Iverson, R.L. 1990. Control of marine fish production. Limnol. Oceanogr. 35:1593-1604. Jamieson, G.S., J. Fulton, D. Mackas, G.A. McFarlane, and D. Ware. 1990. Euphausiid fishery review re potential expansion concerns. Department of Fisheries and Oceans Report, Pacific Biological Station, Nanaimo, BC. 25 p. Jamart, B.M.,D.F. Winter, K. Banse, and R.K. Lam. 1977. A theoretical study of phytoplankton growth and nutrient distribution in the Pacific Ocean off the northwestern U.S. coast. Deep-Sea Res. 24:753-73. 193 Jarre, A., P. Muck, and D. Pauly. 1990. Interactions between fish stocks in the Peruvian upwelling ecosystem. ICES MSM Symposium Paper No. 27. 25 p. Jones, B.C. and G.H. Geen. 1977. Food and feeding of spiny dogfish (Squalus acanthias) in British Columbia waters. J. Fish. Res. Bd Can. 34:2067-2078. Jones, R. 1984. Some observations on energy transfer through the North Sea and Georges Bank food webs. Rapp. P.-v. Reun. Cons. int. Explor. Mer. 183:204-217. Ketchen, K.S. 1986. The spiny dogfish (Squalus acanthias) in the Northeast Pacific and a history of its utilization. Canadian Special Publications of Fisheries and Aquatic Sciences 88:78 p. Kimmerer, W.J. and A.D. McKinnon. 1989. Zooplankton in a marine bay. III. Evidence for influence of vertebrate predation on distributions of two common copepods. Mar. Ecol. Prog. Ser. 53:21-35. Kiorboe, T. 1991. Pelagic fisheries and spatio-temporal variability in zooplankton productivity. Bull. Plank. Soc. Japan. Spec vol (1991):229-249. Kiorboe, T. 1993. Turbulence, phytoplankton cell size, and the structure of pelagic food webs. Advances in Mar. Biol. 29:1-61. Koslow, J.A. 1983. Zooplankton community structure in the North Sea and Northeast Atlantic: Development and test of a biological model. Can. J. Fish. Aquat. Sci. 40:1912-1924. Laevastu, T. F. Favorite, and H.A. Larkins. 1982. Resource assessment and evaluation of the dynamics of the fisheries resources in the northeastern Pacific with numerical ecosystem models, p. 70-81.In M.C.Mercer (ed) Multispecies approaches to fisheries management advice. Can. Spec. Publ. Fish. Aquat. Sci. 59. Landry, M.R. 1981. Switching between herbivory and carnivory by the planktonic marine copepod Calanus pacificus. Mar. Biol. 65:77-82. Landry, M.R., J.R. Postel, W.K.Peterson, and J .Newman. 1989. Broad-scale distributional patterns of hydrographic variables on the Washington/Oregon shelf, p . 1-31.In M.R. Landry and B.M.Hickey (eds) Coastal oceanography of Washington and Oregon. Elsevier Oceanography Serie, Elsevier, New York.. Landry, M.R. and C.J. Lorenzen. 1989. Abundance, distribution, and grazing impact of zooplankton on the Washington shelf .p. 175-202. In M.R.Landry and B.M.Hickey (eds), Coastal oceanography of Washington and Oregon. Elsevier Oceanography Series, Elsevier, New York. Lange, C.B., S.K. Burke, and W. H. Berger. 1990. Biological production off southern California is linked to climate change. Climate Change (June) p. 319-329. Lasker, R. 1966. Feeding, growth, respiration, and carbon utilization of a euphausiid crustacean. J. Fish. Res. Bd. Can. 23:1291-1317. Lasker, R. 1988. Food chains and fisheries: an assessment after 20 years, p. 173-182. In B. Rothschild (ed), Towards a Theory on Biological-Physical Interactions in the World Ocean). Kluwer Academic Publishers. 194 Leaman, B.M. 1992. Groundfish stock assessments for the west coast of Canada in 1991 and recommended yield options for 1992. Can. Tech, Rep. Fish. Aquat. Sci. 1866:304 p. Legendre, L. 1990. The significance of microalgal blooms for fisheries and for the export of particular carbon in oceans. J. Plankton Res. 12:681-699. Legendre, L. and J. Le Fevre. 1992. Interactions between hydrodynamics and pelagic ecosystems: relevance to resource exploitation and climate change. S. Afr. J. mar. Sci. 12:477-486. Lehmann, J.T. 1988. Ecological principles affecting community structure and secondary production by zooplankton in marine and freshwater environments. Limnol. Oceanogr. 33:931-945. Livingston, P.A. 1983. Food habits of Pacific whiting, Merluccius productus, off the west coast of North America, 1967 and 1980. Fish. Bull. 81:629-636. Mackas, D.L. 1992. The seasonal cycle of zooplankton off southwestern British Columbia: 1979-89. Can. J. Fish. Aquat. Sci. 49:1-19. Mackas, D.L. 1994. Interannual variability of the zooplankton community off southern Vancouver Island. Can. J. Fish. Aquat. Sci. 51: In Press. Mackas, D.L. and H.A. Sefton. 1982. Plankton species assemblages off southern Vancouver Island: Geographic pattern and temporal variability. J. mar. Res. 40:1173-1200. Mackas, D.L. and M. Galbraith. 1992. Zooplankton on the west coast of Vancouver Island: distribution and availability to marine birds, p. 15-21.In K. Vermeer, R.W.Butler, and K.H. Morgan (eds), The ecology, status and conservation of marine and shoreline birds on the west coast of Vancouver Island. Can. Wild. Ser, Paper 75. Mackas, D.L., K.L. Denman, and A.F. Bennett. 1987. Least square multiple tracer analysis of water mass composition. J. Geophysical Res. 92:2907-2918. Malone, T.C. 1980. Algal size. p. 433-463. In I. Morris (ed), Physiological Ecology of Phytoplankton. University of California Press, Berkley and Los Angeles. Mann, K.H. 1992. Physical influences on biological processes: How important are they? S. Afr. J. mar. Sci. 12:107-121. Mann, K.H. 1993. Physical oceanography, food chains, and fish stocks: a review. ICES J. mar Sci. 50:105-119. Mann, K.H. and J.R.N. Lazier. 1992. Dynamics of Marine Ecosystems: Biological-Physical Interactions in the Ocean. Boston, Blackwell. 466 p. Mauchline J. 1980. The biology of mysids and euphausiids. Adv. Mar. Biol. 18:595 p. McFarlane, G.A. and R.J. Beamish. 1992. Climate influences linking copepod production with strong year classes in sablefish Anoplopoma fimbria. Can. J. Fish. Aquat. Sci. 49:743-753. McQueen, D.J., J.R. Post, and E.L. Mills. 1986. Trophic relationships in freshwater pelagic ecosystems. Can. J. Fish. Aquat. Sci. 43:1571-1581. 195 McQueen, D.J., M.R.S. Johannes, J.R. Post, T.J.Stewart , and D.R.S. Lean. 1989. Bottom-up and top-down impacts on freshwater pelagic community structure. Ecol. Monogr. 59:289-309. Mills, E X . and R.O. Fournier. 1979. Fish production and the marine ecosystems of the Scotian Shelf, Eastern Canada. Mar. Biol. 54:101-108. Moloney, C.L. 1992. Simulation studies of trophic flows and nutrient cycles in Benguela upwelling foodwebs. S. Afr. J. mar. Sci. 12:457-476. Mullin, M.M.,E.F. Stewart, and F.J. Fuglister. 1975. Ingestion by planktonic grazers as a function of concentration of food. Limnol. Oceanogr. 20:259-262. Mullin, M.M.and A. Conversi. 1988. Biomasses of euphausiids and smaller zooplankton in the California Current: Geographic and interannual comparisons relative to the Pacific Whiting, Merluccius productus. Fish. Bull. 87:633-644. Murphy, G.I. Population biology of the Pacific sardine {Sardinops caerulea). Proceedings of the California Academy of Sciences 4th Series 34:1-84. Nixon, S.W. 1988. Physical energy inputs and the comparative ecology of lake and marine ecosystems. Limnol. Oceanogr. 33:1005-1025. Ohman, M.D. 1984. Omnivory by Euphausia pacifica: the role of copepod prey. Mar. Ecol. Prog. Ser. 19:125-131. Ohman, M.D. 1986. Predator-limited population growth of the copepod Pseudocalanus sp. J. Plankton Res. 8:673-713. Pace, M X . , J.E. Glasser, and L.R. Pomeroy. 1984. A simulation analysis of continental shelf food webs. Mar. Biol. 82:47-63. Paffenhofer, G.A. 1988. Feeding rates and behaviour of zooplankton. Bull. Mar. Sci. 43:430-445. Parsons, T.R. 1979. Some ecological, experimental and evolutionary aspects of the upwelling ecosystem. S. Afr. J. Sci. 75:536-540. Parsons, T.R. 1988. Trophodynamic phasing in theoretical, experimental and natural pelagic ecosystems. J. Oceanogr. Soc. Japan 44:94-101. Parsons, T.R. 1991. The availability of models for determining the effects of climate change and variability on saltwater fisheries of British Columbia. 59 p. In Review of models of climate change and impacts on hydrology, coastal currents and fisheries in British Columbia. Canadian Climate Centre unpublished report No 90-11. Atmospheric Environment Service, National Hydrology Research Centre, Saskatoon, Saskatchewan. Parsons, T.R. and M. Takahashi. 1973. Environmental control of phytoplankton cell size. Limnol Ocean. 18:511-515. Parsons, T.R. and T.A. Kessler. 1987. An ecosystem model for the assessment of plankton production in relation to the survival of young fish. J. Plankton Res. 9:125-137. 196 Parsons, T.R. and C M . Lalli. 1988. Comparative oceanic ecology of the plankton communities of the subarctic Atlantic and Pacific Oceans. Oceanogr. Mar. Biol. Annu. Rev. 26:317-359. Parsons, T.R., R.J. LeBrassuer, and J.D. Fulton. 1967. Some observations on the dependence of zooplankton grazing on the cell size and concentration of phytoplankton blooms. J. Oceanogr. Soc. Japan 23:10-17. Parsons, T.R., M. Takahashi, and B. Hargraves. 1984. Biological Oceanographic Processes. Pergamon Press, New York. 330 p. Perry, M.J. ,J .P. Bolger, and D.C.English. 1989. Primary production in Washington coastal waters, p.117-138. In M.R. Landry and B.M.Hickey (eds), Coastal oceanography of Washington and Oregon. Elsevier Oceanography Series, Elsevier, New York. Peterson, W.T. and C.B.Miller. 1975. Year-to-year variations in the planktology of the Oregon up welling zone. Fish. Bull. 73:642-653. Peterson, W.T., C.B., Miller, and A. Hutchinson. 1979. Zonation and maintenance of copepod populations in the Oregon upwelling zone. Deep-Sea Res. 26:467-494. Pillar, S.C., V. Stuart, M. Barange, and M.J. Gibbons. 1992. Community structure and trophic ecology of euphausiids in the Benguela ecosystem. S. Afr. J. mar. Sci. 12:393-409. Pitcher, G.C., P.C. Brown, and B.A. Mitchell-Innes. 1992. Spatio-temporal variability of phytoplankton in the southern Benguela upwelling system. S. Afr. J. mar. Sci. 12:439-456. Probyn, T.A. 1985. Nitrogen uptake by size-fractionated phytoplankton populations in the southern Benguela upwelling system. Mar. Ecol. Prog. Ser. 22:249-258. Rexstad, E.A. and E.R. Pikitch. 1986. Stomach contents and food consumption estimates of Pacific hake, Merluccius productus. Fish. Bull. 84:947-956. Ritz, D.A. and G.W. Hosie. 1982. Production of the euphausiid Nyctiphanes australis in Storm Bay, south-eastern Tasmania. Mar. Biol. 68:103-108. Roesler, C.S. and D.B. Chelton. 1987. Zooplankton variability in the California Current, 1951-1982. CalCOFI Rep. 28:59-73. Roff, J.C., K. Middlebrook, and F. Evans. 1988. Long-term variability in North Sea zooplankton off the Northumberland coast: Productivity of small copepods and analysis of trophic interactions. J. mar. biol. Ass. U.K. 68:143-161. Ross, R.M. 1982. Energetics of Euphausia pacifica. I. Effects of body carbon and nitrogen and temperature on measured and predicted production. Mar. Biol. 68:1-13. Rothschild, B.J. 1991. On the causes of variability of fish populations - the linkage between large and small scales, p 367-376. In T. Kawasakii, S. Tanaka, Y. Toba, and A. Taniguchi (eds). Long-term variability of pelagic fish populations and their environment, Pergamon Press, Oxford. Rudstram, L.G., S. Hansson, S.Johansson, U. Larsson. 1992. Dynamics of planktivory in a coastal area of the northern Baltic Sea. Mar. Ecol. Prog. Ser. 80:159-173. Ryther, J.H. 1969. Photosynthesis and fish production in the sea. Science 166:72-76. Sameoto, D.D. 1983. Euphausiid distribution in acoustic scattering layers and its significance to surface swarms. J. Plankton Res. 5:129-143. Schweigert, J.F., V. Haist, and C. Fort. 1992. Stock assessment for British Columbia herring in 1992 and forecasts of the potential catch in 1993. PSARC document H92-01 . Shaw, W. R. Tanasichuk, D.M. Ware, D. Davenport, and G.A. McFarlane. 1987. Biological and species interaction survey of Pacific hake, sablefish, Spiny dogfish and Pacific herring off the southwest coast of Vancouver Island. F / V Eastward Ho, August 10-22, 1985. Can. Data Rep. Fish. Aquat. Sci. No. 651: 50 p. Shaw, W. R. Tanasichuk, D.M. Ware, and G.A. McFarlane. 1989a. Biological and species interaction survey of Pacific hake, sablefish, Spiny dogfish and Pacific herring off the southwest coast of Vancouver Island. F /V Caledonian, August 12-25, 1986. Can. MS Rep. Fish. Aquat. Sci. No. 2012: 134 p. Shaw, W. R. Tanasichuk, D.M.Ware, and G.A. McFarlane. 1989b. Biological and species interaction survey of Pacific hake, sablefish, Spiny dogfish and Pacific herring off the southwest coast of Vancouver Island, August 1987. Can. MS Rep. Fish. Aquat. Sci. No. 2027: 149 p. Sheldon, R.W., W.H. Sutcliffe Jr., and M.A. Paranjape. 1977. Structure of pelagic food chain and relationship between plankton and fish production. J. Fish. Res. Bd. Can. 34:2344-2353. Shelton, P.A. 1987. Life-history traits displayed by neritic fish in the Benguela current ecosystem. S. Afr. J. mar. Sci. 5:235-242. Shelton, R.C. 1987. From Hudson Bay to Botany Bay: The lost frigates of Laperouse. NC Press Limited. Toronto. 225 p. Sherman, K. M., Grosslein, D. Mountain, D. Busch, J. O'Reilly, and R. Theroux. 1988. p. 279-337. In Postma, H and J.,J. Zijlstra (eds) Ecosystems of the World: Continental shelves. Elsevier, Amsterdam. Silvert, W.L. 1981. Principles of ecosystem modelling, p. 651-676. In A.R. Longhurst (ed), Analysis of Marine Ecosystem, Academic Press, London. Silvert, W.L. 1989. BSIM software. Habitat Ecology Division, Bedford Institute of Oceanography, Copyright Government of Canada, Halifax, NS. Simpson, J.J. 1992. Response of the southern California current system to the mid-latitude North Pacific coastal warming events of 1982-83 and 1940-41. Fish. Ocean. 1:57-79. Simard, Y. and D.L. Mackas. 1989. Mesoscale aggregations of euphausiid sound scattering layers on the continental shelf of Vancouver Island. Can. J. Fish. Aquat. Sci. 46:1238-1249. Simard, Y. R, de Ladurantaye, J.C. Therriault. 1986. Aggregation of euphausiid along a coastal shelf in an upwelling environment. Mar. Ecol. Prog. Ser. 32:203-215. Smith, P.E. and R.W. Eppley. 1982. Primary production and the anchovy population in the southern California Bight: comparison of time series. Limnol. Oceanogr. 27:1-17. Smith, B.D., G.A. McFarlane, and M.W.Saunders. 1990. Variation in Pacific hake (Merluccius productus) summer length-at-age near southern Vancouver Island and its relationship to fishing and oceanography. Can J. Fish. Aquat. Sci. 47:2195-2211. Smith, P.E., C.H. Lo, J.L. Butler. 1992. Life-stage duration and survival parameters as related to interdecadal population variability in Pacific sardine. CalCOFI Rep. 33:41-49. Sokal, R.R. and F.J. Rohlf. 1981. Biometry. Freeman and Company, New York, 859 p. Strub, P.T. and C. James. 1988. Atmospheric conditions during the spring and autumn transitions in the coastal ocean off western United States. J. Geoph. Res. 93:15561-15584. Stuart, V. 1986. Feeding and metabolism of Euphausia lucens (Euphausiacae) in the southern Benguela current. Mar. Ecol. Prog. Ser. 30:117-125. Stuart, V. and S.C. Pillar. 1988. Growth and production of Euphausia lucens in the southern Benguela Current. J. Plankton Res. 10:1099-1112. Stuart, V. and S.C. Pillar. 1990. Diel patterns of all ontogenetic stages of Euphausia lucens and in situ predation rates on copepods in the southern Benquela upwelling region. Mar. Ecol. Prog. Ser. 64:227-241. Soutar, A. and J.D. Isaacs. 1969. History of fish populations inferred from fish scales in anaerobic sediments off California. CalCOFI Rep. 13:63-70. Swartzman, G.L. and S.P. Kaluzny. 1987. Ecological Simulation Primer. MacMillan Publishing Company. 370 p. Tanasichuk, R.W., D.M. Ware, W.Shaw and G.A. McFarlane. 1991. Variations in diet, daily ration, and feeding periodicity of Pacific hake (Merluccius productus) and spiny dogfish (Squalus acanthias) off the lower west coast of Vancouver Island. Can. J. Fish. Aquat. Sci. 48:2118-2128. Thomas, A.C. and W.J. Emery. 1986. Winter hydrography and plankton distribution on the southern British Columbia continental shelf. Can. J. Fish. Aquat. Sci. 43:1249-1258. Thomson, R.E. and D.M.Ware. 1988. Oceanic factors affecting the distributions and recruitment of west coast fisheries. Can. Tech. Report of Fish, and Aquat. Sci. 1626: 31-64. Thomson, R.E., B.M. Hickey, and P.H. LeBlond. 1989. The Vancouver Island coastal current:fisheries barrier and conduit, p. 265-296. In. R.J. Beamish and G.A. McFarlane (eds), Effects of ocean variability on recruitment and an evaluation of parameters used in stock assessment models. Can. Spec. Publ. Fish. Aquat. Sci. 108. Tont, S.A. 1987. Variability of diatom species populations: from days to years. J. Mar. Res. 45:985-1006. Vennck, E.L., J.A. McGowan, D.R. Cayan, and T.L. Hayward. 1987. Climate and chlorophyll a: long-term trends in the Central North Pacific Ocean. Science 238:70-72. Verheye, H.M., L. Hutchings, J.A. Huggett, and S.J. Painting. 1992. Mesozooplankton dynamics in the Benguela ecosystem, with emphasis on the herbivorous copepods. S. Afr. J. mar. Sci. 12:561-584. Vidal, J. and S.L. Smith. 1986. Biomass, growth, and development of populations of herbivorous zooplankton in the southeastern Bering Sea during spring. Deep-Sea Res. 33:523-556. Wailes, G.H. 1936. Food of Clupea pallasii in southern British Columbia waters. J. Biol. Bd. Can 1:477-486. Walker, D.R. and W.T. Peterson. 1991. Relationships between hydrography, phytoplankton production, biomass, cell size and species composition, and copepod production in the southern Benguela Upwelling System in April 1988. S. Afr. J. mar. Sci. 11:289-305. Walsh, J.J. 1975. A spatial simulation model of the Peru upwelling ecosystem. Deep-sea Res., 22:201-236. Walsh, J.J. 1981. A carbon budget for overfishing off Peru. Nature 290:300-304. Walters, C.J., M. Stocker, A.V. Tyler, and S.J. Westrheim. 1986. Interaction between Pacific cod {Gadus macrocephalus) and herring (Clupea harengus pallasi) in the Hecate Strait, British Columbia. Can. J. Fish. Aquat. Sci. 43:830-837. Ware, D.M. 1982. Power and evolutionary fitness of teleosts. Can. J. Fish. Aquat. Sci. 39:3-13. Ware, D.M. 1991. Climate, predators and prey: behaviour of a linked oscillating system, p. 219-291. In T. Kawasaki (ed), Long-term variability of pelagic fish populations and their environment. Pergamon Press, Oxford. Ware, D.M. 1992. Production characteristics of upwelling systems and the trophodynamic role of hake. S. Afr. J. mar Sci. 12:501-513. Ware, D.M. and R.E. Thomson. 1986. La Perouse Project: First Annual Report, 1985. Dept. Fish. Oceans Report, Nanaimo, B C . 25 p. Ware, D.M. and G.A McFarlane. 1989. Fisheries production domains in the Northeast Pacific Ocean, p . 359-379. In R.J. Beamish and G.A. McFarlane (eds) Effects of ocean variability on recruitment and an evaluation of parameters used in stock assessment models. Can. Spec. Publ. Fish. Aquat. Sci 108. Ware, D.M. and Thomson, R.E. 1991. Link between long-term variability in upwelling and fish production in the northeast pacific Ocean. Can. J. Fish. Aquat. Sci. 48:2296-2306. Ware, D.M. and G.A. McFarlane. 1994. Climate induced changes in hake abundance and pelagic community interactions in the Vancouver Island Upwelling System. Can. Spec. Publ. Fish. Aquat. Sci. 51: In Press. 200 Watkins, L., D.J. Morris, C. Ricketts, and A.W.A. Murray. 1990. Sampling biological characteristics of krill: effect of heterogenous nature of swarms. Mar. Biol. 99:409-416. Wiebe, P.H., S.H. Boyd, B.M.Davis, and J.L. Cox. 1982. Avoidance of towed nets by the euphausiid Nematoscelis mealops. Fish. Bull. 80:75-91. Wilkinson, L. 1990. Systat: The system for statistics. Systat Inc. Evanston, IL. 675 p. Winter, D.F., K. Banse, and G.C. Anderson. 1975. The dynamics of phytoplankton blooms in Puget Sound, a Fjord in the northwestern United States. Mar. Biol. 29:139-176. Wroblewski, J.S. 1982. Interaction of currents and vertical migration in maintaining Calanus marshallae in the Oregon upwelling zone - a simulation. Deep-Sea Res. 29:665-686. Wroblewski, J.S. and J.G. Richman. 1987. The non-linear response of plankton to wind mixing events: implications for survival of larval northern anchovy. J. Plankton Res. 9:103-123. Wyatt, T. 1980. The growth season in the Sea. J. Plankton Res. 2:81-96. Youngbluth, M.J. 1976. Vertical distribution and diel migration of euphausiids in the central region of the California Current. Fish. Bull. 74:925-936. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0053479/manifest

Comment

Related Items